{ "format": "tensorviz.graph", "version": 3, "exportedAt": "2026-09-17T06:02:58.595933+00:00", "status": "static", "warningCount": 0, "document": { "schemaVersion": 2, "semantic": { "graph_id": "snapshot", "name": "SparseRoutingComparison", "revision": 0, "nodes": [ { "node_id": "input", "kind": "input", "op": "Input", "params": { "shape": "1,4,8" }, "in_ports": [], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "input", "name": "Input", "displayName": "Input", "description": "Model input node", "icon": "ArrowRight", "category": "io", "parameters": [ { "name": "shape", "type": "string", "defaultValue": "1,3,224,224", "required": true, "description": "Input tensor shape (comma-separated, e.g., 1,3,224,224)" }, { "name": "name", "type": "string", "defaultValue": "input", "required": false, "description": "Name for this input" } ] } } }, { "node_id": "noisy_topk", "kind": "module", "op": "Module", "params": { "arg_0": 2, "arg_1": true, "arg_2": true }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "moduleInstanceId": "noisy_topk", "moduleTemplateId": "model::SparseExperts", "isGroup": true, "moduleMembers": [ "noisy_topk.router", "noisy_topk.expert0", "noisy_topk.expert1", "add@noisy_topk#0", "noisy_topk.expert2", "add@noisy_topk#1", "noisy_topk.expert3", "add@noisy_topk#2", "noisy_topk.expert4", "add@noisy_topk#3", "noisy_topk.expert5", "add@noisy_topk#4", "noisy_topk.expert6", "add@noisy_topk#5", "noisy_topk.expert7", "add@noisy_topk#6" ], "moduleLayerCount": 95 } }, { "node_id": "switch_top1", "kind": "module", "op": "Module", "params": { "arg_0": 1, "arg_1": false, "arg_2": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "moduleInstanceId": "switch_top1", "moduleTemplateId": "model::SparseExperts", "isGroup": true, "moduleMembers": [ "switch_top1.router", "switch_top1.expert0", "switch_top1.expert1", "add@switch_top1#0", "switch_top1.expert2", "add@switch_top1#1", "switch_top1.expert3", "add@switch_top1#2", "switch_top1.expert4", "add@switch_top1#3", "switch_top1.expert5", "add@switch_top1#4", "switch_top1.expert6", "add@switch_top1#5", "switch_top1.expert7", "add@switch_top1#6" ], "moduleLayerCount": 95 } }, { "node_id": "mixtral_top2", "kind": "module", "op": "Module", "params": { "arg_0": 2, "arg_1": true, "arg_2": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "moduleInstanceId": "mixtral_top2", "moduleTemplateId": "model::SparseExperts", "isGroup": true, "moduleMembers": [ "mixtral_top2.router", "mixtral_top2.expert0", "mixtral_top2.expert1", "add@mixtral_top2#0", "mixtral_top2.expert2", "add@mixtral_top2#1", "mixtral_top2.expert3", "add@mixtral_top2#2", "mixtral_top2.expert4", "add@mixtral_top2#3", "mixtral_top2.expert5", "add@mixtral_top2#4", "mixtral_top2.expert6", "add@mixtral_top2#5", "mixtral_top2.expert7", "add@mixtral_top2#6" ], "moduleLayerCount": 95 } }, { "node_id": "stack@SparseRoutingComparison#0", "kind": "layer", "op": "torch.stack", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "stack", "name": "torch.stack", "displayName": "torch.stack", "description": "PyTorch stack operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] } } }, { "node_id": "output", "kind": "output", "op": "Output", "params": {}, "in_ports": [ "x" ], "out_ports": [], "metadata": { "layerType": { "id": "output", "name": "Output", "displayName": "Output", "description": "Model output node", "icon": "ArrowLeft", "category": "io", "parameters": [ { "name": "name", "type": "string", "defaultValue": "output", "required": false, "description": "Name for this output" } ] } } }, { "node_id": "noisy_topk.router", "kind": "module", "op": "Module", "params": { "arg_0": 2, "arg_1": true, "arg_2": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.router", "moduleTemplateId": "model::Router", "isGroup": true, "moduleMembers": [ "noisy_topk.router.logits", "noisy_topk.router.noise_scale", "softplus@noisy_topk.router#0", "mul@noisy_topk.router#0", "add@noisy_topk.router#0", "python@noisy_topk.router#0", "python@noisy_topk.router#1", "python@noisy_topk.router#2" ], "moduleLayerCount": 8 } }, { "node_id": "noisy_topk.expert0", "kind": "module", "op": "Module", "params": { "arg_0": 0 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert0", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert0#0", "python@noisy_topk.expert0#0", "python@noisy_topk.expert0#1", "noisy_topk.expert0.expert", "mul@noisy_topk.expert0#0", "python@noisy_topk.expert0#2" ], "moduleLayerCount": 10 } }, { "node_id": "noisy_topk.expert1", "kind": "module", "op": "Module", "params": { "arg_0": 1 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert1", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert1#0", "python@noisy_topk.expert1#0", "python@noisy_topk.expert1#1", "noisy_topk.expert1.expert", "mul@noisy_topk.expert1#0", "python@noisy_topk.expert1#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@noisy_topk#0", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk" } }, { "node_id": "noisy_topk.expert2", "kind": "module", "op": "Module", "params": { "arg_0": 2 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert2", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert2#0", "python@noisy_topk.expert2#0", "python@noisy_topk.expert2#1", "noisy_topk.expert2.expert", "mul@noisy_topk.expert2#0", "python@noisy_topk.expert2#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@noisy_topk#1", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk" } }, { "node_id": "noisy_topk.expert3", "kind": "module", "op": "Module", "params": { "arg_0": 3 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert3", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert3#0", "python@noisy_topk.expert3#0", "python@noisy_topk.expert3#1", "noisy_topk.expert3.expert", "mul@noisy_topk.expert3#0", "python@noisy_topk.expert3#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@noisy_topk#2", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk" } }, { "node_id": "noisy_topk.expert4", "kind": "module", "op": "Module", "params": { "arg_0": 4 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert4", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert4#0", "python@noisy_topk.expert4#0", "python@noisy_topk.expert4#1", "noisy_topk.expert4.expert", "mul@noisy_topk.expert4#0", "python@noisy_topk.expert4#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@noisy_topk#3", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk" } }, { "node_id": "noisy_topk.expert5", "kind": "module", "op": "Module", "params": { "arg_0": 5 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert5", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert5#0", "python@noisy_topk.expert5#0", "python@noisy_topk.expert5#1", "noisy_topk.expert5.expert", "mul@noisy_topk.expert5#0", "python@noisy_topk.expert5#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@noisy_topk#4", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk" } }, { "node_id": "noisy_topk.expert6", "kind": "module", "op": "Module", "params": { "arg_0": 6 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert6", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert6#0", "python@noisy_topk.expert6#0", "python@noisy_topk.expert6#1", "noisy_topk.expert6.expert", "mul@noisy_topk.expert6#0", "python@noisy_topk.expert6#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@noisy_topk#5", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk" } }, { "node_id": "noisy_topk.expert7", "kind": "module", "op": "Module", "params": { "arg_0": 7 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk", "moduleInstanceId": "noisy_topk.expert7", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@noisy_topk.expert7#0", "python@noisy_topk.expert7#0", "python@noisy_topk.expert7#1", "noisy_topk.expert7.expert", "mul@noisy_topk.expert7#0", "python@noisy_topk.expert7#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@noisy_topk#6", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk" } }, { "node_id": "noisy_topk.router.logits", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.router" } }, { "node_id": "noisy_topk.router.noise_scale", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.router" } }, { "node_id": "softplus@noisy_topk.router#0", "kind": "layer", "op": "Softplus", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softplus", "name": "Softplus", "displayName": "Softplus", "description": "Softplus activation: (1/beta) * log(1 + exp(beta * x))", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "beta", "type": "number", "defaultValue": 1, "required": false, "description": "The beta value for the Softplus formulation", "min": 0 }, { "name": "threshold", "type": "number", "defaultValue": 20, "required": false, "description": "Values above this revert to a linear function", "min": 0 } ] }, "parentId": "noisy_topk.router" } }, { "node_id": "mul@noisy_topk.router#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.router" } }, { "node_id": "add@noisy_topk.router#0", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.router" } }, { "node_id": "python@noisy_topk.router#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.router" } }, { "node_id": "python@noisy_topk.router#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.router" } }, { "node_id": "python@noisy_topk.router#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.router" } }, { "node_id": "flatten@noisy_topk.expert0#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert0" } }, { "node_id": "python@noisy_topk.expert0#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert0" } }, { "node_id": "python@noisy_topk.expert0#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert0" } }, { "node_id": "noisy_topk.expert0.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert0", "moduleInstanceId": "noisy_topk.expert0.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert0.expert.gate", "noisy_topk.expert0.expert.up", "noisy_topk.expert0.expert.silu", "mul@noisy_topk.expert0.expert#0", "noisy_topk.expert0.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert0#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert0" } }, { "node_id": "python@noisy_topk.expert0#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert0" } }, { "node_id": "noisy_topk.expert0.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert0.expert" } }, { "node_id": "noisy_topk.expert0.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert0.expert" } }, { "node_id": "noisy_topk.expert0.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert0.expert" } }, { "node_id": "mul@noisy_topk.expert0.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert0.expert" } }, { "node_id": "noisy_topk.expert0.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert0.expert" } }, { "node_id": "flatten@noisy_topk.expert1#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert1" } }, { "node_id": "python@noisy_topk.expert1#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert1" } }, { "node_id": "python@noisy_topk.expert1#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert1" } }, { "node_id": "noisy_topk.expert1.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert1", "moduleInstanceId": "noisy_topk.expert1.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert1.expert.gate", "noisy_topk.expert1.expert.up", "noisy_topk.expert1.expert.silu", "mul@noisy_topk.expert1.expert#0", "noisy_topk.expert1.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert1#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert1" } }, { "node_id": "python@noisy_topk.expert1#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert1" } }, { "node_id": "noisy_topk.expert1.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert1.expert" } }, { "node_id": "noisy_topk.expert1.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert1.expert" } }, { "node_id": "noisy_topk.expert1.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert1.expert" } }, { "node_id": "mul@noisy_topk.expert1.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert1.expert" } }, { "node_id": "noisy_topk.expert1.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert1.expert" } }, { "node_id": "flatten@noisy_topk.expert2#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert2" } }, { "node_id": "python@noisy_topk.expert2#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert2" } }, { "node_id": "python@noisy_topk.expert2#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert2" } }, { "node_id": "noisy_topk.expert2.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert2", "moduleInstanceId": "noisy_topk.expert2.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert2.expert.gate", "noisy_topk.expert2.expert.up", "noisy_topk.expert2.expert.silu", "mul@noisy_topk.expert2.expert#0", "noisy_topk.expert2.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert2#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert2" } }, { "node_id": "python@noisy_topk.expert2#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert2" } }, { "node_id": "noisy_topk.expert2.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert2.expert" } }, { "node_id": "noisy_topk.expert2.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert2.expert" } }, { "node_id": "noisy_topk.expert2.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert2.expert" } }, { "node_id": "mul@noisy_topk.expert2.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert2.expert" } }, { "node_id": "noisy_topk.expert2.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert2.expert" } }, { "node_id": "flatten@noisy_topk.expert3#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert3" } }, { "node_id": "python@noisy_topk.expert3#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert3" } }, { "node_id": "python@noisy_topk.expert3#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert3" } }, { "node_id": "noisy_topk.expert3.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert3", "moduleInstanceId": "noisy_topk.expert3.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert3.expert.gate", "noisy_topk.expert3.expert.up", "noisy_topk.expert3.expert.silu", "mul@noisy_topk.expert3.expert#0", "noisy_topk.expert3.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert3#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert3" } }, { "node_id": "python@noisy_topk.expert3#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert3" } }, { "node_id": "noisy_topk.expert3.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert3.expert" } }, { "node_id": "noisy_topk.expert3.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert3.expert" } }, { "node_id": "noisy_topk.expert3.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert3.expert" } }, { "node_id": "mul@noisy_topk.expert3.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert3.expert" } }, { "node_id": "noisy_topk.expert3.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert3.expert" } }, { "node_id": "flatten@noisy_topk.expert4#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert4" } }, { "node_id": "python@noisy_topk.expert4#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert4" } }, { "node_id": "python@noisy_topk.expert4#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert4" } }, { "node_id": "noisy_topk.expert4.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert4", "moduleInstanceId": "noisy_topk.expert4.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert4.expert.gate", "noisy_topk.expert4.expert.up", "noisy_topk.expert4.expert.silu", "mul@noisy_topk.expert4.expert#0", "noisy_topk.expert4.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert4#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert4" } }, { "node_id": "python@noisy_topk.expert4#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert4" } }, { "node_id": "noisy_topk.expert4.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert4.expert" } }, { "node_id": "noisy_topk.expert4.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert4.expert" } }, { "node_id": "noisy_topk.expert4.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert4.expert" } }, { "node_id": "mul@noisy_topk.expert4.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert4.expert" } }, { "node_id": "noisy_topk.expert4.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert4.expert" } }, { "node_id": "flatten@noisy_topk.expert5#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert5" } }, { "node_id": "python@noisy_topk.expert5#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert5" } }, { "node_id": "python@noisy_topk.expert5#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert5" } }, { "node_id": "noisy_topk.expert5.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert5", "moduleInstanceId": "noisy_topk.expert5.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert5.expert.gate", "noisy_topk.expert5.expert.up", "noisy_topk.expert5.expert.silu", "mul@noisy_topk.expert5.expert#0", "noisy_topk.expert5.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert5#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert5" } }, { "node_id": "python@noisy_topk.expert5#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert5" } }, { "node_id": "noisy_topk.expert5.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert5.expert" } }, { "node_id": "noisy_topk.expert5.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert5.expert" } }, { "node_id": "noisy_topk.expert5.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert5.expert" } }, { "node_id": "mul@noisy_topk.expert5.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert5.expert" } }, { "node_id": "noisy_topk.expert5.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert5.expert" } }, { "node_id": "flatten@noisy_topk.expert6#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert6" } }, { "node_id": "python@noisy_topk.expert6#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert6" } }, { "node_id": "python@noisy_topk.expert6#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert6" } }, { "node_id": "noisy_topk.expert6.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert6", "moduleInstanceId": "noisy_topk.expert6.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert6.expert.gate", "noisy_topk.expert6.expert.up", "noisy_topk.expert6.expert.silu", "mul@noisy_topk.expert6.expert#0", "noisy_topk.expert6.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert6#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert6" } }, { "node_id": "python@noisy_topk.expert6#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert6" } }, { "node_id": "noisy_topk.expert6.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert6.expert" } }, { "node_id": "noisy_topk.expert6.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert6.expert" } }, { "node_id": "noisy_topk.expert6.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert6.expert" } }, { "node_id": "mul@noisy_topk.expert6.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert6.expert" } }, { "node_id": "noisy_topk.expert6.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert6.expert" } }, { "node_id": "flatten@noisy_topk.expert7#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "noisy_topk.expert7" } }, { "node_id": "python@noisy_topk.expert7#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert7" } }, { "node_id": "python@noisy_topk.expert7#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert7" } }, { "node_id": "noisy_topk.expert7.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "noisy_topk.expert7", "moduleInstanceId": "noisy_topk.expert7.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "noisy_topk.expert7.expert.gate", "noisy_topk.expert7.expert.up", "noisy_topk.expert7.expert.silu", "mul@noisy_topk.expert7.expert#0", "noisy_topk.expert7.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@noisy_topk.expert7#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert7" } }, { "node_id": "python@noisy_topk.expert7#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "noisy_topk.expert7" } }, { "node_id": "noisy_topk.expert7.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert7.expert" } }, { "node_id": "noisy_topk.expert7.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert7.expert" } }, { "node_id": "noisy_topk.expert7.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "noisy_topk.expert7.expert" } }, { "node_id": "mul@noisy_topk.expert7.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "noisy_topk.expert7.expert" } }, { "node_id": "noisy_topk.expert7.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "noisy_topk.expert7.expert" } }, { "node_id": "switch_top1.router", "kind": "module", "op": "Module", "params": { "arg_0": 2, "arg_1": true, "arg_2": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.router", "moduleTemplateId": "model::Router", "isGroup": true, "moduleMembers": [ "switch_top1.router.logits", "switch_top1.router.noise_scale", "softplus@switch_top1.router#0", "mul@switch_top1.router#0", "add@switch_top1.router#0", "python@switch_top1.router#0", "python@switch_top1.router#1", "python@switch_top1.router#2" ], "moduleLayerCount": 8 } }, { "node_id": "switch_top1.expert0", "kind": "module", "op": "Module", "params": { "arg_0": 0 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert0", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert0#0", "python@switch_top1.expert0#0", "python@switch_top1.expert0#1", "switch_top1.expert0.expert", "mul@switch_top1.expert0#0", "python@switch_top1.expert0#2" ], "moduleLayerCount": 10 } }, { "node_id": "switch_top1.expert1", "kind": "module", "op": "Module", "params": { "arg_0": 1 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert1", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert1#0", "python@switch_top1.expert1#0", "python@switch_top1.expert1#1", "switch_top1.expert1.expert", "mul@switch_top1.expert1#0", "python@switch_top1.expert1#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@switch_top1#0", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1" } }, { "node_id": "switch_top1.expert2", "kind": "module", "op": "Module", "params": { "arg_0": 2 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert2", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert2#0", "python@switch_top1.expert2#0", "python@switch_top1.expert2#1", "switch_top1.expert2.expert", "mul@switch_top1.expert2#0", "python@switch_top1.expert2#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@switch_top1#1", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1" } }, { "node_id": "switch_top1.expert3", "kind": "module", "op": "Module", "params": { "arg_0": 3 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert3", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert3#0", "python@switch_top1.expert3#0", "python@switch_top1.expert3#1", "switch_top1.expert3.expert", "mul@switch_top1.expert3#0", "python@switch_top1.expert3#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@switch_top1#2", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1" } }, { "node_id": "switch_top1.expert4", "kind": "module", "op": "Module", "params": { "arg_0": 4 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert4", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert4#0", "python@switch_top1.expert4#0", "python@switch_top1.expert4#1", "switch_top1.expert4.expert", "mul@switch_top1.expert4#0", "python@switch_top1.expert4#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@switch_top1#3", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1" } }, { "node_id": "switch_top1.expert5", "kind": "module", "op": "Module", "params": { "arg_0": 5 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert5", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert5#0", "python@switch_top1.expert5#0", "python@switch_top1.expert5#1", "switch_top1.expert5.expert", "mul@switch_top1.expert5#0", "python@switch_top1.expert5#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@switch_top1#4", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1" } }, { "node_id": "switch_top1.expert6", "kind": "module", "op": "Module", "params": { "arg_0": 6 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert6", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert6#0", "python@switch_top1.expert6#0", "python@switch_top1.expert6#1", "switch_top1.expert6.expert", "mul@switch_top1.expert6#0", "python@switch_top1.expert6#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@switch_top1#5", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1" } }, { "node_id": "switch_top1.expert7", "kind": "module", "op": "Module", "params": { "arg_0": 7 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1", "moduleInstanceId": "switch_top1.expert7", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@switch_top1.expert7#0", "python@switch_top1.expert7#0", "python@switch_top1.expert7#1", "switch_top1.expert7.expert", "mul@switch_top1.expert7#0", "python@switch_top1.expert7#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@switch_top1#6", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1" } }, { "node_id": "switch_top1.router.logits", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.router" } }, { "node_id": "switch_top1.router.noise_scale", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.router" } }, { "node_id": "softplus@switch_top1.router#0", "kind": "layer", "op": "Softplus", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softplus", "name": "Softplus", "displayName": "Softplus", "description": "Softplus activation: (1/beta) * log(1 + exp(beta * x))", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "beta", "type": "number", "defaultValue": 1, "required": false, "description": "The beta value for the Softplus formulation", "min": 0 }, { "name": "threshold", "type": "number", "defaultValue": 20, "required": false, "description": "Values above this revert to a linear function", "min": 0 } ] }, "parentId": "switch_top1.router" } }, { "node_id": "mul@switch_top1.router#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.router" } }, { "node_id": "add@switch_top1.router#0", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.router" } }, { "node_id": "python@switch_top1.router#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.router" } }, { "node_id": "python@switch_top1.router#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.router" } }, { "node_id": "python@switch_top1.router#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.router" } }, { "node_id": "flatten@switch_top1.expert0#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert0" } }, { "node_id": "python@switch_top1.expert0#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert0" } }, { "node_id": "python@switch_top1.expert0#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert0" } }, { "node_id": "switch_top1.expert0.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert0", "moduleInstanceId": "switch_top1.expert0.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert0.expert.gate", "switch_top1.expert0.expert.up", "switch_top1.expert0.expert.silu", "mul@switch_top1.expert0.expert#0", "switch_top1.expert0.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert0#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert0" } }, { "node_id": "python@switch_top1.expert0#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert0" } }, { "node_id": "switch_top1.expert0.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert0.expert" } }, { "node_id": "switch_top1.expert0.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert0.expert" } }, { "node_id": "switch_top1.expert0.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert0.expert" } }, { "node_id": "mul@switch_top1.expert0.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert0.expert" } }, { "node_id": "switch_top1.expert0.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert0.expert" } }, { "node_id": "flatten@switch_top1.expert1#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert1" } }, { "node_id": "python@switch_top1.expert1#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert1" } }, { "node_id": "python@switch_top1.expert1#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert1" } }, { "node_id": "switch_top1.expert1.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert1", "moduleInstanceId": "switch_top1.expert1.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert1.expert.gate", "switch_top1.expert1.expert.up", "switch_top1.expert1.expert.silu", "mul@switch_top1.expert1.expert#0", "switch_top1.expert1.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert1#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert1" } }, { "node_id": "python@switch_top1.expert1#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert1" } }, { "node_id": "switch_top1.expert1.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert1.expert" } }, { "node_id": "switch_top1.expert1.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert1.expert" } }, { "node_id": "switch_top1.expert1.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert1.expert" } }, { "node_id": "mul@switch_top1.expert1.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert1.expert" } }, { "node_id": "switch_top1.expert1.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert1.expert" } }, { "node_id": "flatten@switch_top1.expert2#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert2" } }, { "node_id": "python@switch_top1.expert2#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert2" } }, { "node_id": "python@switch_top1.expert2#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert2" } }, { "node_id": "switch_top1.expert2.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert2", "moduleInstanceId": "switch_top1.expert2.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert2.expert.gate", "switch_top1.expert2.expert.up", "switch_top1.expert2.expert.silu", "mul@switch_top1.expert2.expert#0", "switch_top1.expert2.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert2#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert2" } }, { "node_id": "python@switch_top1.expert2#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert2" } }, { "node_id": "switch_top1.expert2.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert2.expert" } }, { "node_id": "switch_top1.expert2.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert2.expert" } }, { "node_id": "switch_top1.expert2.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert2.expert" } }, { "node_id": "mul@switch_top1.expert2.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert2.expert" } }, { "node_id": "switch_top1.expert2.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert2.expert" } }, { "node_id": "flatten@switch_top1.expert3#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert3" } }, { "node_id": "python@switch_top1.expert3#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert3" } }, { "node_id": "python@switch_top1.expert3#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert3" } }, { "node_id": "switch_top1.expert3.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert3", "moduleInstanceId": "switch_top1.expert3.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert3.expert.gate", "switch_top1.expert3.expert.up", "switch_top1.expert3.expert.silu", "mul@switch_top1.expert3.expert#0", "switch_top1.expert3.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert3#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert3" } }, { "node_id": "python@switch_top1.expert3#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert3" } }, { "node_id": "switch_top1.expert3.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert3.expert" } }, { "node_id": "switch_top1.expert3.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert3.expert" } }, { "node_id": "switch_top1.expert3.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert3.expert" } }, { "node_id": "mul@switch_top1.expert3.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert3.expert" } }, { "node_id": "switch_top1.expert3.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert3.expert" } }, { "node_id": "flatten@switch_top1.expert4#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert4" } }, { "node_id": "python@switch_top1.expert4#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert4" } }, { "node_id": "python@switch_top1.expert4#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert4" } }, { "node_id": "switch_top1.expert4.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert4", "moduleInstanceId": "switch_top1.expert4.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert4.expert.gate", "switch_top1.expert4.expert.up", "switch_top1.expert4.expert.silu", "mul@switch_top1.expert4.expert#0", "switch_top1.expert4.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert4#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert4" } }, { "node_id": "python@switch_top1.expert4#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert4" } }, { "node_id": "switch_top1.expert4.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert4.expert" } }, { "node_id": "switch_top1.expert4.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert4.expert" } }, { "node_id": "switch_top1.expert4.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert4.expert" } }, { "node_id": "mul@switch_top1.expert4.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert4.expert" } }, { "node_id": "switch_top1.expert4.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert4.expert" } }, { "node_id": "flatten@switch_top1.expert5#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert5" } }, { "node_id": "python@switch_top1.expert5#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert5" } }, { "node_id": "python@switch_top1.expert5#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert5" } }, { "node_id": "switch_top1.expert5.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert5", "moduleInstanceId": "switch_top1.expert5.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert5.expert.gate", "switch_top1.expert5.expert.up", "switch_top1.expert5.expert.silu", "mul@switch_top1.expert5.expert#0", "switch_top1.expert5.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert5#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert5" } }, { "node_id": "python@switch_top1.expert5#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert5" } }, { "node_id": "switch_top1.expert5.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert5.expert" } }, { "node_id": "switch_top1.expert5.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert5.expert" } }, { "node_id": "switch_top1.expert5.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert5.expert" } }, { "node_id": "mul@switch_top1.expert5.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert5.expert" } }, { "node_id": "switch_top1.expert5.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert5.expert" } }, { "node_id": "flatten@switch_top1.expert6#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert6" } }, { "node_id": "python@switch_top1.expert6#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert6" } }, { "node_id": "python@switch_top1.expert6#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert6" } }, { "node_id": "switch_top1.expert6.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert6", "moduleInstanceId": "switch_top1.expert6.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert6.expert.gate", "switch_top1.expert6.expert.up", "switch_top1.expert6.expert.silu", "mul@switch_top1.expert6.expert#0", "switch_top1.expert6.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert6#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert6" } }, { "node_id": "python@switch_top1.expert6#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert6" } }, { "node_id": "switch_top1.expert6.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert6.expert" } }, { "node_id": "switch_top1.expert6.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert6.expert" } }, { "node_id": "switch_top1.expert6.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert6.expert" } }, { "node_id": "mul@switch_top1.expert6.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert6.expert" } }, { "node_id": "switch_top1.expert6.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert6.expert" } }, { "node_id": "flatten@switch_top1.expert7#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "switch_top1.expert7" } }, { "node_id": "python@switch_top1.expert7#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert7" } }, { "node_id": "python@switch_top1.expert7#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert7" } }, { "node_id": "switch_top1.expert7.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "switch_top1.expert7", "moduleInstanceId": "switch_top1.expert7.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "switch_top1.expert7.expert.gate", "switch_top1.expert7.expert.up", "switch_top1.expert7.expert.silu", "mul@switch_top1.expert7.expert#0", "switch_top1.expert7.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@switch_top1.expert7#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert7" } }, { "node_id": "python@switch_top1.expert7#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "switch_top1.expert7" } }, { "node_id": "switch_top1.expert7.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert7.expert" } }, { "node_id": "switch_top1.expert7.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert7.expert" } }, { "node_id": "switch_top1.expert7.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "switch_top1.expert7.expert" } }, { "node_id": "mul@switch_top1.expert7.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "switch_top1.expert7.expert" } }, { "node_id": "switch_top1.expert7.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "switch_top1.expert7.expert" } }, { "node_id": "mixtral_top2.router", "kind": "module", "op": "Module", "params": { "arg_0": 2, "arg_1": true, "arg_2": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.router", "moduleTemplateId": "model::Router", "isGroup": true, "moduleMembers": [ "mixtral_top2.router.logits", "mixtral_top2.router.noise_scale", "softplus@mixtral_top2.router#0", "mul@mixtral_top2.router#0", "add@mixtral_top2.router#0", "python@mixtral_top2.router#0", "python@mixtral_top2.router#1", "python@mixtral_top2.router#2" ], "moduleLayerCount": 8 } }, { "node_id": "mixtral_top2.expert0", "kind": "module", "op": "Module", "params": { "arg_0": 0 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert0", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert0#0", "python@mixtral_top2.expert0#0", "python@mixtral_top2.expert0#1", "mixtral_top2.expert0.expert", "mul@mixtral_top2.expert0#0", "python@mixtral_top2.expert0#2" ], "moduleLayerCount": 10 } }, { "node_id": "mixtral_top2.expert1", "kind": "module", "op": "Module", "params": { "arg_0": 1 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert1", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert1#0", "python@mixtral_top2.expert1#0", "python@mixtral_top2.expert1#1", "mixtral_top2.expert1.expert", "mul@mixtral_top2.expert1#0", "python@mixtral_top2.expert1#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@mixtral_top2#0", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2" } }, { "node_id": "mixtral_top2.expert2", "kind": "module", "op": "Module", "params": { "arg_0": 2 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert2", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert2#0", "python@mixtral_top2.expert2#0", "python@mixtral_top2.expert2#1", "mixtral_top2.expert2.expert", "mul@mixtral_top2.expert2#0", "python@mixtral_top2.expert2#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@mixtral_top2#1", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2" } }, { "node_id": "mixtral_top2.expert3", "kind": "module", "op": "Module", "params": { "arg_0": 3 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert3", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert3#0", "python@mixtral_top2.expert3#0", "python@mixtral_top2.expert3#1", "mixtral_top2.expert3.expert", "mul@mixtral_top2.expert3#0", "python@mixtral_top2.expert3#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@mixtral_top2#2", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2" } }, { "node_id": "mixtral_top2.expert4", "kind": "module", "op": "Module", "params": { "arg_0": 4 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert4", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert4#0", "python@mixtral_top2.expert4#0", "python@mixtral_top2.expert4#1", "mixtral_top2.expert4.expert", "mul@mixtral_top2.expert4#0", "python@mixtral_top2.expert4#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@mixtral_top2#3", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2" } }, { "node_id": "mixtral_top2.expert5", "kind": "module", "op": "Module", "params": { "arg_0": 5 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert5", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert5#0", "python@mixtral_top2.expert5#0", "python@mixtral_top2.expert5#1", "mixtral_top2.expert5.expert", "mul@mixtral_top2.expert5#0", "python@mixtral_top2.expert5#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@mixtral_top2#4", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2" } }, { "node_id": "mixtral_top2.expert6", "kind": "module", "op": "Module", "params": { "arg_0": 6 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert6", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert6#0", "python@mixtral_top2.expert6#0", "python@mixtral_top2.expert6#1", "mixtral_top2.expert6.expert", "mul@mixtral_top2.expert6#0", "python@mixtral_top2.expert6#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@mixtral_top2#5", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2" } }, { "node_id": "mixtral_top2.expert7", "kind": "module", "op": "Module", "params": { "arg_0": 7 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2", "moduleInstanceId": "mixtral_top2.expert7", "moduleTemplateId": "model::Dispatch", "isGroup": true, "moduleMembers": [ "flatten@mixtral_top2.expert7#0", "python@mixtral_top2.expert7#0", "python@mixtral_top2.expert7#1", "mixtral_top2.expert7.expert", "mul@mixtral_top2.expert7#0", "python@mixtral_top2.expert7#2" ], "moduleLayerCount": 10 } }, { "node_id": "add@mixtral_top2#6", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2" } }, { "node_id": "mixtral_top2.router.logits", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.router" } }, { "node_id": "mixtral_top2.router.noise_scale", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.router" } }, { "node_id": "softplus@mixtral_top2.router#0", "kind": "layer", "op": "Softplus", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softplus", "name": "Softplus", "displayName": "Softplus", "description": "Softplus activation: (1/beta) * log(1 + exp(beta * x))", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "beta", "type": "number", "defaultValue": 1, "required": false, "description": "The beta value for the Softplus formulation", "min": 0 }, { "name": "threshold", "type": "number", "defaultValue": 20, "required": false, "description": "Values above this revert to a linear function", "min": 0 } ] }, "parentId": "mixtral_top2.router" } }, { "node_id": "mul@mixtral_top2.router#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.router" } }, { "node_id": "add@mixtral_top2.router#0", "kind": "layer", "op": "torch.add", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "add", "name": "torch.add", "displayName": "torch.add", "description": "PyTorch add operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.router" } }, { "node_id": "python@mixtral_top2.router#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.router" } }, { "node_id": "python@mixtral_top2.router#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.router" } }, { "node_id": "python@mixtral_top2.router#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.router" } }, { "node_id": "flatten@mixtral_top2.expert0#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert0" } }, { "node_id": "python@mixtral_top2.expert0#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert0" } }, { "node_id": "python@mixtral_top2.expert0#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert0" } }, { "node_id": "mixtral_top2.expert0.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert0", "moduleInstanceId": "mixtral_top2.expert0.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert0.expert.gate", "mixtral_top2.expert0.expert.up", "mixtral_top2.expert0.expert.silu", "mul@mixtral_top2.expert0.expert#0", "mixtral_top2.expert0.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert0#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert0" } }, { "node_id": "python@mixtral_top2.expert0#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert0" } }, { "node_id": "mixtral_top2.expert0.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert0.expert" } }, { "node_id": "mixtral_top2.expert0.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert0.expert" } }, { "node_id": "mixtral_top2.expert0.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert0.expert" } }, { "node_id": "mul@mixtral_top2.expert0.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert0.expert" } }, { "node_id": "mixtral_top2.expert0.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert0.expert" } }, { "node_id": "flatten@mixtral_top2.expert1#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert1" } }, { "node_id": "python@mixtral_top2.expert1#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert1" } }, { "node_id": "python@mixtral_top2.expert1#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert1" } }, { "node_id": "mixtral_top2.expert1.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert1", "moduleInstanceId": "mixtral_top2.expert1.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert1.expert.gate", "mixtral_top2.expert1.expert.up", "mixtral_top2.expert1.expert.silu", "mul@mixtral_top2.expert1.expert#0", "mixtral_top2.expert1.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert1#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert1" } }, { "node_id": "python@mixtral_top2.expert1#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert1" } }, { "node_id": "mixtral_top2.expert1.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert1.expert" } }, { "node_id": "mixtral_top2.expert1.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert1.expert" } }, { "node_id": "mixtral_top2.expert1.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert1.expert" } }, { "node_id": "mul@mixtral_top2.expert1.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert1.expert" } }, { "node_id": "mixtral_top2.expert1.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert1.expert" } }, { "node_id": "flatten@mixtral_top2.expert2#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert2" } }, { "node_id": "python@mixtral_top2.expert2#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert2" } }, { "node_id": "python@mixtral_top2.expert2#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert2" } }, { "node_id": "mixtral_top2.expert2.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert2", "moduleInstanceId": "mixtral_top2.expert2.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert2.expert.gate", "mixtral_top2.expert2.expert.up", "mixtral_top2.expert2.expert.silu", "mul@mixtral_top2.expert2.expert#0", "mixtral_top2.expert2.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert2#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert2" } }, { "node_id": "python@mixtral_top2.expert2#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert2" } }, { "node_id": "mixtral_top2.expert2.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert2.expert" } }, { "node_id": "mixtral_top2.expert2.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert2.expert" } }, { "node_id": "mixtral_top2.expert2.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert2.expert" } }, { "node_id": "mul@mixtral_top2.expert2.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert2.expert" } }, { "node_id": "mixtral_top2.expert2.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert2.expert" } }, { "node_id": "flatten@mixtral_top2.expert3#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert3" } }, { "node_id": "python@mixtral_top2.expert3#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert3" } }, { "node_id": "python@mixtral_top2.expert3#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert3" } }, { "node_id": "mixtral_top2.expert3.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert3", "moduleInstanceId": "mixtral_top2.expert3.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert3.expert.gate", "mixtral_top2.expert3.expert.up", "mixtral_top2.expert3.expert.silu", "mul@mixtral_top2.expert3.expert#0", "mixtral_top2.expert3.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert3#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert3" } }, { "node_id": "python@mixtral_top2.expert3#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert3" } }, { "node_id": "mixtral_top2.expert3.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert3.expert" } }, { "node_id": "mixtral_top2.expert3.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert3.expert" } }, { "node_id": "mixtral_top2.expert3.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert3.expert" } }, { "node_id": "mul@mixtral_top2.expert3.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert3.expert" } }, { "node_id": "mixtral_top2.expert3.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert3.expert" } }, { "node_id": "flatten@mixtral_top2.expert4#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert4" } }, { "node_id": "python@mixtral_top2.expert4#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert4" } }, { "node_id": "python@mixtral_top2.expert4#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert4" } }, { "node_id": "mixtral_top2.expert4.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert4", "moduleInstanceId": "mixtral_top2.expert4.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert4.expert.gate", "mixtral_top2.expert4.expert.up", "mixtral_top2.expert4.expert.silu", "mul@mixtral_top2.expert4.expert#0", "mixtral_top2.expert4.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert4#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert4" } }, { "node_id": "python@mixtral_top2.expert4#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert4" } }, { "node_id": "mixtral_top2.expert4.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert4.expert" } }, { "node_id": "mixtral_top2.expert4.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert4.expert" } }, { "node_id": "mixtral_top2.expert4.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert4.expert" } }, { "node_id": "mul@mixtral_top2.expert4.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert4.expert" } }, { "node_id": "mixtral_top2.expert4.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert4.expert" } }, { "node_id": "flatten@mixtral_top2.expert5#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert5" } }, { "node_id": "python@mixtral_top2.expert5#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert5" } }, { "node_id": "python@mixtral_top2.expert5#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert5" } }, { "node_id": "mixtral_top2.expert5.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert5", "moduleInstanceId": "mixtral_top2.expert5.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert5.expert.gate", "mixtral_top2.expert5.expert.up", "mixtral_top2.expert5.expert.silu", "mul@mixtral_top2.expert5.expert#0", "mixtral_top2.expert5.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert5#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert5" } }, { "node_id": "python@mixtral_top2.expert5#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert5" } }, { "node_id": "mixtral_top2.expert5.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert5.expert" } }, { "node_id": "mixtral_top2.expert5.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert5.expert" } }, { "node_id": "mixtral_top2.expert5.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert5.expert" } }, { "node_id": "mul@mixtral_top2.expert5.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert5.expert" } }, { "node_id": "mixtral_top2.expert5.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert5.expert" } }, { "node_id": "flatten@mixtral_top2.expert6#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert6" } }, { "node_id": "python@mixtral_top2.expert6#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert6" } }, { "node_id": "python@mixtral_top2.expert6#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert6" } }, { "node_id": "mixtral_top2.expert6.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert6", "moduleInstanceId": "mixtral_top2.expert6.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert6.expert.gate", "mixtral_top2.expert6.expert.up", "mixtral_top2.expert6.expert.silu", "mul@mixtral_top2.expert6.expert#0", "mixtral_top2.expert6.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert6#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert6" } }, { "node_id": "python@mixtral_top2.expert6#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert6" } }, { "node_id": "mixtral_top2.expert6.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert6.expert" } }, { "node_id": "mixtral_top2.expert6.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert6.expert" } }, { "node_id": "mixtral_top2.expert6.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert6.expert" } }, { "node_id": "mul@mixtral_top2.expert6.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert6.expert" } }, { "node_id": "mixtral_top2.expert6.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert6.expert" } }, { "node_id": "flatten@mixtral_top2.expert7#0", "kind": "layer", "op": "Flatten", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "flatten", "name": "Flatten", "displayName": "Flatten", "description": "Flattens input tensor contiguously", "icon": "Maximize2", "category": "tensor", "parameters": [ { "name": "start_dim", "type": "number", "defaultValue": 1, "required": false, "description": "First dim to flatten", "min": 0 }, { "name": "end_dim", "type": "number", "defaultValue": -1, "required": false, "description": "Last dim to flatten" } ] }, "parentId": "mixtral_top2.expert7" } }, { "node_id": "python@mixtral_top2.expert7#0", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert7" } }, { "node_id": "python@mixtral_top2.expert7#1", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert7" } }, { "node_id": "mixtral_top2.expert7.expert", "kind": "module", "op": "Module", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "module", "name": "Module", "displayName": "Module", "description": "Custom module group (nn.Module)", "icon": "Package", "category": "module", "parameters": [ { "name": "name", "type": "string", "defaultValue": "CustomModule", "required": true, "description": "Module class name" }, { "name": "description", "type": "string", "defaultValue": "", "required": false, "description": "Module description" } ] }, "parentId": "mixtral_top2.expert7", "moduleInstanceId": "mixtral_top2.expert7.expert", "moduleTemplateId": "model::Expert", "isGroup": true, "moduleMembers": [ "mixtral_top2.expert7.expert.gate", "mixtral_top2.expert7.expert.up", "mixtral_top2.expert7.expert.silu", "mul@mixtral_top2.expert7.expert#0", "mixtral_top2.expert7.expert.down" ], "moduleLayerCount": 5 } }, { "node_id": "mul@mixtral_top2.expert7#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert7" } }, { "node_id": "python@mixtral_top2.expert7#2", "kind": "layer", "op": "Python", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "python", "name": "Python", "displayName": "Python", "description": "A statement from the source file that has no graph equivalent. Its code is kept verbatim and its shape is unknown until the model is traced.", "icon": "Hash", "category": "misc", "parameters": [] }, "parentId": "mixtral_top2.expert7" } }, { "node_id": "mixtral_top2.expert7.expert.gate", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert7.expert" } }, { "node_id": "mixtral_top2.expert7.expert.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 12, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert7.expert" } }, { "node_id": "mixtral_top2.expert7.expert.silu", "kind": "layer", "op": "SiLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "silu", "name": "SiLU", "displayName": "SiLU", "description": "Sigmoid Linear Unit (Swish)", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "mixtral_top2.expert7.expert" } }, { "node_id": "mul@mixtral_top2.expert7.expert#0", "kind": "layer", "op": "torch.mul", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "mul", "name": "torch.mul", "displayName": "torch.mul", "description": "PyTorch mul operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "mixtral_top2.expert7.expert" } }, { "node_id": "mixtral_top2.expert7.expert.down", "kind": "layer", "op": "Linear", "params": { "in_features": 12, "out_features": 8, "bias": false }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "linear", "name": "Linear", "displayName": "Linear", "description": "Fully Connected Layer", "icon": "Minus", "category": "linear", "parameters": [ { "name": "in_features", "type": "number", "defaultValue": 512, "required": true, "description": "Size of input features", "min": 1 }, { "name": "out_features", "type": "number", "defaultValue": 256, "required": true, "description": "Size of output features", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If set to False, the layer will not learn an additive bias" } ] }, "parentId": "mixtral_top2.expert7.expert" } } ], "edges": [ { "edge_id": "edge-1", "source_node_id": "noisy_topk.router.noise_scale", "target_node_id": "softplus@noisy_topk.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-2", "source_node_id": "softplus@noisy_topk.router#0", "target_node_id": "mul@noisy_topk.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-3", "source_node_id": "noisy_topk.router.logits", "target_node_id": "mul@noisy_topk.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-4", "source_node_id": "noisy_topk.router.logits", "target_node_id": "add@noisy_topk.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-5", "source_node_id": "mul@noisy_topk.router#0", "target_node_id": "add@noisy_topk.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-6", "source_node_id": "add@noisy_topk.router#0", "target_node_id": "python@noisy_topk.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-7", "source_node_id": "python@noisy_topk.router#0", "target_node_id": "python@noisy_topk.router#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-8", "source_node_id": "python@noisy_topk.router#0", "target_node_id": "python@noisy_topk.router#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-9", "source_node_id": "python@noisy_topk.router#1", "target_node_id": "python@noisy_topk.router#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-10", "source_node_id": "noisy_topk.expert0.expert.gate", "target_node_id": "noisy_topk.expert0.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-11", "source_node_id": "noisy_topk.expert0.expert.silu", "target_node_id": "mul@noisy_topk.expert0.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-12", "source_node_id": "noisy_topk.expert0.expert.up", "target_node_id": "mul@noisy_topk.expert0.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-13", "source_node_id": "mul@noisy_topk.expert0.expert#0", "target_node_id": "noisy_topk.expert0.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-14", "source_node_id": "python@noisy_topk.expert0#0", "target_node_id": "python@noisy_topk.expert0#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-15", "source_node_id": "flatten@noisy_topk.expert0#0", "target_node_id": "python@noisy_topk.expert0#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-16", "source_node_id": "python@noisy_topk.expert0#1", "target_node_id": "noisy_topk.expert0.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-17", "source_node_id": "python@noisy_topk.expert0#1", "target_node_id": "noisy_topk.expert0.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-18", "source_node_id": "python@noisy_topk.expert0#1", "target_node_id": "noisy_topk.expert0.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-19", "source_node_id": "noisy_topk.expert0.expert.down", "target_node_id": "mul@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-20", "source_node_id": "noisy_topk.expert0.expert", "target_node_id": "mul@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-21", "source_node_id": "flatten@noisy_topk.expert0#0", "target_node_id": "mul@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-22", "source_node_id": "python@noisy_topk.expert0#0", "target_node_id": "python@noisy_topk.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-23", "source_node_id": "flatten@noisy_topk.expert0#0", "target_node_id": "python@noisy_topk.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-24", "source_node_id": "mul@noisy_topk.expert0#0", "target_node_id": "python@noisy_topk.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-25", "source_node_id": "noisy_topk.expert1.expert.gate", "target_node_id": "noisy_topk.expert1.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-26", "source_node_id": "noisy_topk.expert1.expert.silu", "target_node_id": "mul@noisy_topk.expert1.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-27", "source_node_id": "noisy_topk.expert1.expert.up", "target_node_id": "mul@noisy_topk.expert1.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-28", "source_node_id": "mul@noisy_topk.expert1.expert#0", "target_node_id": "noisy_topk.expert1.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-29", "source_node_id": "python@noisy_topk.expert1#0", "target_node_id": "python@noisy_topk.expert1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-30", "source_node_id": "flatten@noisy_topk.expert1#0", "target_node_id": "python@noisy_topk.expert1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-31", "source_node_id": "python@noisy_topk.expert1#1", "target_node_id": "noisy_topk.expert1.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-32", "source_node_id": "python@noisy_topk.expert1#1", "target_node_id": "noisy_topk.expert1.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-33", "source_node_id": "python@noisy_topk.expert1#1", "target_node_id": "noisy_topk.expert1.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-34", "source_node_id": "noisy_topk.expert1.expert.down", "target_node_id": "mul@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-35", "source_node_id": "noisy_topk.expert1.expert", "target_node_id": "mul@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-36", "source_node_id": "flatten@noisy_topk.expert1#0", "target_node_id": "mul@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-37", "source_node_id": "python@noisy_topk.expert1#0", "target_node_id": "python@noisy_topk.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-38", "source_node_id": "flatten@noisy_topk.expert1#0", "target_node_id": "python@noisy_topk.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-39", "source_node_id": "mul@noisy_topk.expert1#0", "target_node_id": "python@noisy_topk.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-40", "source_node_id": "noisy_topk.expert2.expert.gate", "target_node_id": "noisy_topk.expert2.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-41", "source_node_id": "noisy_topk.expert2.expert.silu", "target_node_id": "mul@noisy_topk.expert2.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-42", "source_node_id": "noisy_topk.expert2.expert.up", "target_node_id": "mul@noisy_topk.expert2.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-43", "source_node_id": "mul@noisy_topk.expert2.expert#0", "target_node_id": "noisy_topk.expert2.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-44", "source_node_id": "python@noisy_topk.expert2#0", "target_node_id": "python@noisy_topk.expert2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-45", "source_node_id": "flatten@noisy_topk.expert2#0", "target_node_id": "python@noisy_topk.expert2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-46", "source_node_id": "python@noisy_topk.expert2#1", "target_node_id": "noisy_topk.expert2.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-47", "source_node_id": "python@noisy_topk.expert2#1", "target_node_id": "noisy_topk.expert2.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-48", "source_node_id": "python@noisy_topk.expert2#1", "target_node_id": "noisy_topk.expert2.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-49", "source_node_id": "noisy_topk.expert2.expert.down", "target_node_id": "mul@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-50", "source_node_id": "noisy_topk.expert2.expert", "target_node_id": "mul@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-51", "source_node_id": "flatten@noisy_topk.expert2#0", "target_node_id": "mul@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-52", "source_node_id": "python@noisy_topk.expert2#0", "target_node_id": "python@noisy_topk.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-53", "source_node_id": "flatten@noisy_topk.expert2#0", "target_node_id": "python@noisy_topk.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-54", "source_node_id": "mul@noisy_topk.expert2#0", "target_node_id": "python@noisy_topk.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-55", "source_node_id": "noisy_topk.expert3.expert.gate", "target_node_id": "noisy_topk.expert3.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-56", "source_node_id": "noisy_topk.expert3.expert.silu", "target_node_id": "mul@noisy_topk.expert3.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-57", "source_node_id": "noisy_topk.expert3.expert.up", "target_node_id": "mul@noisy_topk.expert3.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-58", "source_node_id": "mul@noisy_topk.expert3.expert#0", "target_node_id": "noisy_topk.expert3.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-59", "source_node_id": "python@noisy_topk.expert3#0", "target_node_id": "python@noisy_topk.expert3#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-60", "source_node_id": "flatten@noisy_topk.expert3#0", "target_node_id": "python@noisy_topk.expert3#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-61", "source_node_id": "python@noisy_topk.expert3#1", "target_node_id": "noisy_topk.expert3.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-62", "source_node_id": "python@noisy_topk.expert3#1", "target_node_id": "noisy_topk.expert3.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-63", "source_node_id": "python@noisy_topk.expert3#1", "target_node_id": "noisy_topk.expert3.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-64", "source_node_id": "noisy_topk.expert3.expert.down", "target_node_id": "mul@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-65", "source_node_id": "noisy_topk.expert3.expert", "target_node_id": "mul@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-66", "source_node_id": "flatten@noisy_topk.expert3#0", "target_node_id": "mul@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-67", "source_node_id": "python@noisy_topk.expert3#0", "target_node_id": "python@noisy_topk.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-68", "source_node_id": "flatten@noisy_topk.expert3#0", "target_node_id": "python@noisy_topk.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-69", "source_node_id": "mul@noisy_topk.expert3#0", "target_node_id": "python@noisy_topk.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-70", "source_node_id": "noisy_topk.expert4.expert.gate", "target_node_id": "noisy_topk.expert4.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-71", "source_node_id": "noisy_topk.expert4.expert.silu", "target_node_id": "mul@noisy_topk.expert4.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-72", "source_node_id": "noisy_topk.expert4.expert.up", "target_node_id": "mul@noisy_topk.expert4.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-73", "source_node_id": "mul@noisy_topk.expert4.expert#0", "target_node_id": "noisy_topk.expert4.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-74", "source_node_id": "python@noisy_topk.expert4#0", "target_node_id": "python@noisy_topk.expert4#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-75", "source_node_id": "flatten@noisy_topk.expert4#0", "target_node_id": "python@noisy_topk.expert4#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-76", "source_node_id": "python@noisy_topk.expert4#1", "target_node_id": "noisy_topk.expert4.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-77", "source_node_id": "python@noisy_topk.expert4#1", "target_node_id": "noisy_topk.expert4.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-78", "source_node_id": "python@noisy_topk.expert4#1", "target_node_id": "noisy_topk.expert4.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-79", "source_node_id": "noisy_topk.expert4.expert.down", "target_node_id": "mul@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-80", "source_node_id": "noisy_topk.expert4.expert", "target_node_id": "mul@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-81", "source_node_id": "flatten@noisy_topk.expert4#0", "target_node_id": "mul@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-82", "source_node_id": "python@noisy_topk.expert4#0", "target_node_id": "python@noisy_topk.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-83", "source_node_id": "flatten@noisy_topk.expert4#0", "target_node_id": "python@noisy_topk.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-84", "source_node_id": "mul@noisy_topk.expert4#0", "target_node_id": "python@noisy_topk.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-85", "source_node_id": "noisy_topk.expert5.expert.gate", "target_node_id": "noisy_topk.expert5.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-86", "source_node_id": "noisy_topk.expert5.expert.silu", "target_node_id": "mul@noisy_topk.expert5.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-87", "source_node_id": "noisy_topk.expert5.expert.up", "target_node_id": "mul@noisy_topk.expert5.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-88", "source_node_id": "mul@noisy_topk.expert5.expert#0", "target_node_id": "noisy_topk.expert5.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-89", "source_node_id": "python@noisy_topk.expert5#0", "target_node_id": "python@noisy_topk.expert5#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-90", "source_node_id": "flatten@noisy_topk.expert5#0", "target_node_id": "python@noisy_topk.expert5#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-91", "source_node_id": "python@noisy_topk.expert5#1", "target_node_id": "noisy_topk.expert5.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-92", "source_node_id": "python@noisy_topk.expert5#1", "target_node_id": "noisy_topk.expert5.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-93", "source_node_id": "python@noisy_topk.expert5#1", "target_node_id": "noisy_topk.expert5.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-94", "source_node_id": "noisy_topk.expert5.expert.down", "target_node_id": "mul@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-95", "source_node_id": "noisy_topk.expert5.expert", "target_node_id": "mul@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-96", "source_node_id": "flatten@noisy_topk.expert5#0", "target_node_id": "mul@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-97", "source_node_id": "python@noisy_topk.expert5#0", "target_node_id": "python@noisy_topk.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-98", "source_node_id": "flatten@noisy_topk.expert5#0", "target_node_id": "python@noisy_topk.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-99", "source_node_id": "mul@noisy_topk.expert5#0", "target_node_id": "python@noisy_topk.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-100", "source_node_id": "noisy_topk.expert6.expert.gate", "target_node_id": "noisy_topk.expert6.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-101", "source_node_id": "noisy_topk.expert6.expert.silu", "target_node_id": "mul@noisy_topk.expert6.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-102", "source_node_id": "noisy_topk.expert6.expert.up", "target_node_id": "mul@noisy_topk.expert6.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-103", "source_node_id": "mul@noisy_topk.expert6.expert#0", "target_node_id": "noisy_topk.expert6.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-104", "source_node_id": "python@noisy_topk.expert6#0", "target_node_id": "python@noisy_topk.expert6#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-105", "source_node_id": "flatten@noisy_topk.expert6#0", "target_node_id": "python@noisy_topk.expert6#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-106", "source_node_id": "python@noisy_topk.expert6#1", "target_node_id": "noisy_topk.expert6.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-107", "source_node_id": "python@noisy_topk.expert6#1", "target_node_id": "noisy_topk.expert6.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-108", "source_node_id": "python@noisy_topk.expert6#1", "target_node_id": "noisy_topk.expert6.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-109", "source_node_id": "noisy_topk.expert6.expert.down", "target_node_id": "mul@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-110", "source_node_id": "noisy_topk.expert6.expert", "target_node_id": "mul@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-111", "source_node_id": "flatten@noisy_topk.expert6#0", "target_node_id": "mul@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-112", "source_node_id": "python@noisy_topk.expert6#0", "target_node_id": "python@noisy_topk.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-113", "source_node_id": "flatten@noisy_topk.expert6#0", "target_node_id": "python@noisy_topk.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-114", "source_node_id": "mul@noisy_topk.expert6#0", "target_node_id": "python@noisy_topk.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-115", "source_node_id": "noisy_topk.expert7.expert.gate", "target_node_id": "noisy_topk.expert7.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-116", "source_node_id": "noisy_topk.expert7.expert.silu", "target_node_id": "mul@noisy_topk.expert7.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-117", "source_node_id": "noisy_topk.expert7.expert.up", "target_node_id": "mul@noisy_topk.expert7.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-118", "source_node_id": "mul@noisy_topk.expert7.expert#0", "target_node_id": "noisy_topk.expert7.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-119", "source_node_id": "python@noisy_topk.expert7#0", "target_node_id": "python@noisy_topk.expert7#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-120", "source_node_id": "flatten@noisy_topk.expert7#0", "target_node_id": "python@noisy_topk.expert7#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-121", "source_node_id": "python@noisy_topk.expert7#1", "target_node_id": "noisy_topk.expert7.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-122", "source_node_id": "python@noisy_topk.expert7#1", "target_node_id": "noisy_topk.expert7.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-123", "source_node_id": "python@noisy_topk.expert7#1", "target_node_id": "noisy_topk.expert7.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-124", "source_node_id": "noisy_topk.expert7.expert.down", "target_node_id": "mul@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-125", "source_node_id": "noisy_topk.expert7.expert", "target_node_id": "mul@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-126", "source_node_id": "flatten@noisy_topk.expert7#0", "target_node_id": "mul@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-127", "source_node_id": "python@noisy_topk.expert7#0", "target_node_id": "python@noisy_topk.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-128", "source_node_id": "flatten@noisy_topk.expert7#0", "target_node_id": "python@noisy_topk.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-129", "source_node_id": "mul@noisy_topk.expert7#0", "target_node_id": "python@noisy_topk.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-130", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-131", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-132", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-133", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-134", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-135", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-136", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-137", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-138", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-139", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-140", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-141", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-142", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-143", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-144", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-145", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-146", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-147", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-148", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-149", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-150", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-151", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-152", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-153", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-154", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-155", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-156", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-157", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-158", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-159", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-160", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-161", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-162", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-163", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-164", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-165", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-166", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "flatten@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-167", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "mul@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-168", "source_node_id": "python@noisy_topk.router#2", "target_node_id": "python@noisy_topk.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-169", "source_node_id": "noisy_topk.router", "target_node_id": "noisy_topk.expert7", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-170", "source_node_id": "python@noisy_topk.expert0#2", "target_node_id": "add@noisy_topk#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-171", "source_node_id": "noisy_topk.expert0", "target_node_id": "add@noisy_topk#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-172", "source_node_id": "python@noisy_topk.expert1#2", "target_node_id": "add@noisy_topk#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-173", "source_node_id": "noisy_topk.expert1", "target_node_id": "add@noisy_topk#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-174", "source_node_id": "add@noisy_topk#0", "target_node_id": "add@noisy_topk#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-175", "source_node_id": "python@noisy_topk.expert2#2", "target_node_id": "add@noisy_topk#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-176", "source_node_id": "noisy_topk.expert2", "target_node_id": "add@noisy_topk#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-177", "source_node_id": "add@noisy_topk#1", "target_node_id": "add@noisy_topk#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-178", "source_node_id": "python@noisy_topk.expert3#2", "target_node_id": "add@noisy_topk#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-179", "source_node_id": "noisy_topk.expert3", "target_node_id": "add@noisy_topk#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-180", "source_node_id": "add@noisy_topk#2", "target_node_id": "add@noisy_topk#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-181", "source_node_id": "python@noisy_topk.expert4#2", "target_node_id": "add@noisy_topk#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-182", "source_node_id": "noisy_topk.expert4", "target_node_id": "add@noisy_topk#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-183", "source_node_id": "add@noisy_topk#3", "target_node_id": "add@noisy_topk#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-184", "source_node_id": "python@noisy_topk.expert5#2", "target_node_id": "add@noisy_topk#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-185", "source_node_id": "noisy_topk.expert5", "target_node_id": "add@noisy_topk#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-186", "source_node_id": "add@noisy_topk#4", "target_node_id": "add@noisy_topk#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-187", "source_node_id": "python@noisy_topk.expert6#2", "target_node_id": "add@noisy_topk#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-188", "source_node_id": "noisy_topk.expert6", "target_node_id": "add@noisy_topk#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-189", "source_node_id": "add@noisy_topk#5", "target_node_id": "add@noisy_topk#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-190", "source_node_id": "python@noisy_topk.expert7#2", "target_node_id": "add@noisy_topk#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-191", "source_node_id": "noisy_topk.expert7", "target_node_id": "add@noisy_topk#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-192", "source_node_id": "switch_top1.router.noise_scale", "target_node_id": "softplus@switch_top1.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-193", "source_node_id": "softplus@switch_top1.router#0", "target_node_id": "mul@switch_top1.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-194", "source_node_id": "switch_top1.router.logits", "target_node_id": "mul@switch_top1.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-195", "source_node_id": "switch_top1.router.logits", "target_node_id": "add@switch_top1.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-196", "source_node_id": "mul@switch_top1.router#0", "target_node_id": "add@switch_top1.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-197", "source_node_id": "add@switch_top1.router#0", "target_node_id": "python@switch_top1.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-198", "source_node_id": "python@switch_top1.router#0", "target_node_id": "python@switch_top1.router#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-199", "source_node_id": "python@switch_top1.router#0", "target_node_id": "python@switch_top1.router#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-200", "source_node_id": "python@switch_top1.router#1", "target_node_id": "python@switch_top1.router#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-201", "source_node_id": "switch_top1.expert0.expert.gate", "target_node_id": "switch_top1.expert0.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-202", "source_node_id": "switch_top1.expert0.expert.silu", "target_node_id": "mul@switch_top1.expert0.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-203", "source_node_id": "switch_top1.expert0.expert.up", "target_node_id": "mul@switch_top1.expert0.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-204", "source_node_id": "mul@switch_top1.expert0.expert#0", "target_node_id": "switch_top1.expert0.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-205", "source_node_id": "python@switch_top1.expert0#0", "target_node_id": "python@switch_top1.expert0#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-206", "source_node_id": "flatten@switch_top1.expert0#0", "target_node_id": "python@switch_top1.expert0#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-207", "source_node_id": "python@switch_top1.expert0#1", "target_node_id": "switch_top1.expert0.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-208", "source_node_id": "python@switch_top1.expert0#1", "target_node_id": "switch_top1.expert0.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-209", "source_node_id": "python@switch_top1.expert0#1", "target_node_id": "switch_top1.expert0.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-210", "source_node_id": "switch_top1.expert0.expert.down", "target_node_id": "mul@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-211", "source_node_id": "switch_top1.expert0.expert", "target_node_id": "mul@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-212", "source_node_id": "flatten@switch_top1.expert0#0", "target_node_id": "mul@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-213", "source_node_id": "python@switch_top1.expert0#0", "target_node_id": "python@switch_top1.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-214", "source_node_id": "flatten@switch_top1.expert0#0", "target_node_id": "python@switch_top1.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-215", "source_node_id": "mul@switch_top1.expert0#0", "target_node_id": "python@switch_top1.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-216", "source_node_id": "switch_top1.expert1.expert.gate", "target_node_id": "switch_top1.expert1.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-217", "source_node_id": "switch_top1.expert1.expert.silu", "target_node_id": "mul@switch_top1.expert1.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-218", "source_node_id": "switch_top1.expert1.expert.up", "target_node_id": "mul@switch_top1.expert1.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-219", "source_node_id": "mul@switch_top1.expert1.expert#0", "target_node_id": "switch_top1.expert1.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-220", "source_node_id": "python@switch_top1.expert1#0", "target_node_id": "python@switch_top1.expert1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-221", "source_node_id": "flatten@switch_top1.expert1#0", "target_node_id": "python@switch_top1.expert1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-222", "source_node_id": "python@switch_top1.expert1#1", "target_node_id": "switch_top1.expert1.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-223", "source_node_id": "python@switch_top1.expert1#1", "target_node_id": "switch_top1.expert1.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-224", "source_node_id": "python@switch_top1.expert1#1", "target_node_id": "switch_top1.expert1.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-225", "source_node_id": "switch_top1.expert1.expert.down", "target_node_id": "mul@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-226", "source_node_id": "switch_top1.expert1.expert", "target_node_id": "mul@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-227", "source_node_id": "flatten@switch_top1.expert1#0", "target_node_id": "mul@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-228", "source_node_id": "python@switch_top1.expert1#0", "target_node_id": "python@switch_top1.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-229", "source_node_id": "flatten@switch_top1.expert1#0", "target_node_id": "python@switch_top1.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-230", "source_node_id": "mul@switch_top1.expert1#0", "target_node_id": "python@switch_top1.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-231", "source_node_id": "switch_top1.expert2.expert.gate", "target_node_id": "switch_top1.expert2.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-232", "source_node_id": "switch_top1.expert2.expert.silu", "target_node_id": "mul@switch_top1.expert2.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-233", "source_node_id": "switch_top1.expert2.expert.up", "target_node_id": "mul@switch_top1.expert2.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-234", "source_node_id": "mul@switch_top1.expert2.expert#0", "target_node_id": "switch_top1.expert2.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-235", "source_node_id": "python@switch_top1.expert2#0", "target_node_id": "python@switch_top1.expert2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-236", "source_node_id": "flatten@switch_top1.expert2#0", "target_node_id": "python@switch_top1.expert2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-237", "source_node_id": "python@switch_top1.expert2#1", "target_node_id": "switch_top1.expert2.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-238", "source_node_id": "python@switch_top1.expert2#1", "target_node_id": "switch_top1.expert2.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-239", "source_node_id": "python@switch_top1.expert2#1", "target_node_id": "switch_top1.expert2.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-240", "source_node_id": "switch_top1.expert2.expert.down", "target_node_id": "mul@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-241", "source_node_id": "switch_top1.expert2.expert", "target_node_id": "mul@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-242", "source_node_id": "flatten@switch_top1.expert2#0", "target_node_id": "mul@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-243", "source_node_id": "python@switch_top1.expert2#0", "target_node_id": "python@switch_top1.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-244", "source_node_id": "flatten@switch_top1.expert2#0", "target_node_id": "python@switch_top1.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-245", "source_node_id": "mul@switch_top1.expert2#0", "target_node_id": "python@switch_top1.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-246", "source_node_id": "switch_top1.expert3.expert.gate", "target_node_id": "switch_top1.expert3.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-247", "source_node_id": "switch_top1.expert3.expert.silu", "target_node_id": "mul@switch_top1.expert3.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-248", "source_node_id": "switch_top1.expert3.expert.up", "target_node_id": "mul@switch_top1.expert3.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-249", "source_node_id": "mul@switch_top1.expert3.expert#0", "target_node_id": "switch_top1.expert3.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-250", "source_node_id": "python@switch_top1.expert3#0", "target_node_id": "python@switch_top1.expert3#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-251", "source_node_id": "flatten@switch_top1.expert3#0", "target_node_id": "python@switch_top1.expert3#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-252", "source_node_id": "python@switch_top1.expert3#1", "target_node_id": "switch_top1.expert3.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-253", "source_node_id": "python@switch_top1.expert3#1", "target_node_id": "switch_top1.expert3.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-254", "source_node_id": "python@switch_top1.expert3#1", "target_node_id": "switch_top1.expert3.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-255", "source_node_id": "switch_top1.expert3.expert.down", "target_node_id": "mul@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-256", "source_node_id": "switch_top1.expert3.expert", "target_node_id": "mul@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-257", "source_node_id": "flatten@switch_top1.expert3#0", "target_node_id": "mul@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-258", "source_node_id": "python@switch_top1.expert3#0", "target_node_id": "python@switch_top1.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-259", "source_node_id": "flatten@switch_top1.expert3#0", "target_node_id": "python@switch_top1.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-260", "source_node_id": "mul@switch_top1.expert3#0", "target_node_id": "python@switch_top1.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-261", "source_node_id": "switch_top1.expert4.expert.gate", "target_node_id": "switch_top1.expert4.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-262", "source_node_id": "switch_top1.expert4.expert.silu", "target_node_id": "mul@switch_top1.expert4.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-263", "source_node_id": "switch_top1.expert4.expert.up", "target_node_id": "mul@switch_top1.expert4.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-264", "source_node_id": "mul@switch_top1.expert4.expert#0", "target_node_id": "switch_top1.expert4.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-265", "source_node_id": "python@switch_top1.expert4#0", "target_node_id": "python@switch_top1.expert4#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-266", "source_node_id": "flatten@switch_top1.expert4#0", "target_node_id": "python@switch_top1.expert4#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-267", "source_node_id": "python@switch_top1.expert4#1", "target_node_id": "switch_top1.expert4.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-268", "source_node_id": "python@switch_top1.expert4#1", "target_node_id": "switch_top1.expert4.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-269", "source_node_id": "python@switch_top1.expert4#1", "target_node_id": "switch_top1.expert4.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-270", "source_node_id": "switch_top1.expert4.expert.down", "target_node_id": "mul@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-271", "source_node_id": "switch_top1.expert4.expert", "target_node_id": "mul@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-272", "source_node_id": "flatten@switch_top1.expert4#0", "target_node_id": "mul@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-273", "source_node_id": "python@switch_top1.expert4#0", "target_node_id": "python@switch_top1.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-274", "source_node_id": "flatten@switch_top1.expert4#0", "target_node_id": "python@switch_top1.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-275", "source_node_id": "mul@switch_top1.expert4#0", "target_node_id": "python@switch_top1.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-276", "source_node_id": "switch_top1.expert5.expert.gate", "target_node_id": "switch_top1.expert5.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-277", "source_node_id": "switch_top1.expert5.expert.silu", "target_node_id": "mul@switch_top1.expert5.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-278", "source_node_id": "switch_top1.expert5.expert.up", "target_node_id": "mul@switch_top1.expert5.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-279", "source_node_id": "mul@switch_top1.expert5.expert#0", "target_node_id": "switch_top1.expert5.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-280", "source_node_id": "python@switch_top1.expert5#0", "target_node_id": "python@switch_top1.expert5#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-281", "source_node_id": "flatten@switch_top1.expert5#0", "target_node_id": "python@switch_top1.expert5#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-282", "source_node_id": "python@switch_top1.expert5#1", "target_node_id": "switch_top1.expert5.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-283", "source_node_id": "python@switch_top1.expert5#1", "target_node_id": "switch_top1.expert5.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-284", "source_node_id": "python@switch_top1.expert5#1", "target_node_id": "switch_top1.expert5.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-285", "source_node_id": "switch_top1.expert5.expert.down", "target_node_id": "mul@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-286", "source_node_id": "switch_top1.expert5.expert", "target_node_id": "mul@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-287", "source_node_id": "flatten@switch_top1.expert5#0", "target_node_id": "mul@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-288", "source_node_id": "python@switch_top1.expert5#0", "target_node_id": "python@switch_top1.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-289", "source_node_id": "flatten@switch_top1.expert5#0", "target_node_id": "python@switch_top1.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-290", "source_node_id": "mul@switch_top1.expert5#0", "target_node_id": "python@switch_top1.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-291", "source_node_id": "switch_top1.expert6.expert.gate", "target_node_id": "switch_top1.expert6.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-292", "source_node_id": "switch_top1.expert6.expert.silu", "target_node_id": "mul@switch_top1.expert6.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-293", "source_node_id": "switch_top1.expert6.expert.up", "target_node_id": "mul@switch_top1.expert6.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-294", "source_node_id": "mul@switch_top1.expert6.expert#0", "target_node_id": "switch_top1.expert6.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-295", "source_node_id": "python@switch_top1.expert6#0", "target_node_id": "python@switch_top1.expert6#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-296", "source_node_id": "flatten@switch_top1.expert6#0", "target_node_id": "python@switch_top1.expert6#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-297", "source_node_id": "python@switch_top1.expert6#1", "target_node_id": "switch_top1.expert6.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-298", "source_node_id": "python@switch_top1.expert6#1", "target_node_id": "switch_top1.expert6.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-299", "source_node_id": "python@switch_top1.expert6#1", "target_node_id": "switch_top1.expert6.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-300", "source_node_id": "switch_top1.expert6.expert.down", "target_node_id": "mul@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-301", "source_node_id": "switch_top1.expert6.expert", "target_node_id": "mul@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-302", "source_node_id": "flatten@switch_top1.expert6#0", "target_node_id": "mul@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-303", "source_node_id": "python@switch_top1.expert6#0", "target_node_id": "python@switch_top1.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-304", "source_node_id": "flatten@switch_top1.expert6#0", "target_node_id": "python@switch_top1.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-305", "source_node_id": "mul@switch_top1.expert6#0", "target_node_id": "python@switch_top1.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-306", "source_node_id": "switch_top1.expert7.expert.gate", "target_node_id": "switch_top1.expert7.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-307", "source_node_id": "switch_top1.expert7.expert.silu", "target_node_id": "mul@switch_top1.expert7.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-308", "source_node_id": "switch_top1.expert7.expert.up", "target_node_id": "mul@switch_top1.expert7.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-309", "source_node_id": "mul@switch_top1.expert7.expert#0", "target_node_id": "switch_top1.expert7.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-310", "source_node_id": "python@switch_top1.expert7#0", "target_node_id": "python@switch_top1.expert7#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-311", "source_node_id": "flatten@switch_top1.expert7#0", "target_node_id": "python@switch_top1.expert7#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-312", "source_node_id": "python@switch_top1.expert7#1", "target_node_id": "switch_top1.expert7.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-313", "source_node_id": "python@switch_top1.expert7#1", "target_node_id": "switch_top1.expert7.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-314", "source_node_id": "python@switch_top1.expert7#1", "target_node_id": "switch_top1.expert7.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-315", "source_node_id": "switch_top1.expert7.expert.down", "target_node_id": "mul@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-316", "source_node_id": "switch_top1.expert7.expert", "target_node_id": "mul@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-317", "source_node_id": "flatten@switch_top1.expert7#0", "target_node_id": "mul@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-318", "source_node_id": "python@switch_top1.expert7#0", "target_node_id": "python@switch_top1.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-319", "source_node_id": "flatten@switch_top1.expert7#0", "target_node_id": "python@switch_top1.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-320", "source_node_id": "mul@switch_top1.expert7#0", "target_node_id": "python@switch_top1.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-321", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-322", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-323", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-324", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-325", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-326", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-327", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-328", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-329", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-330", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-331", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-332", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-333", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-334", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-335", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-336", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-337", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-338", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-339", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-340", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-341", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-342", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-343", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-344", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-345", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-346", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-347", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-348", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-349", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-350", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-351", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-352", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-353", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-354", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-355", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-356", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-357", "source_node_id": "python@switch_top1.router#2", "target_node_id": "flatten@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-358", "source_node_id": "python@switch_top1.router#2", "target_node_id": "mul@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-359", "source_node_id": "python@switch_top1.router#2", "target_node_id": "python@switch_top1.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-360", "source_node_id": "switch_top1.router", "target_node_id": "switch_top1.expert7", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-361", "source_node_id": "python@switch_top1.expert0#2", "target_node_id": "add@switch_top1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-362", "source_node_id": "switch_top1.expert0", "target_node_id": "add@switch_top1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-363", "source_node_id": "python@switch_top1.expert1#2", "target_node_id": "add@switch_top1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-364", "source_node_id": "switch_top1.expert1", "target_node_id": "add@switch_top1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-365", "source_node_id": "add@switch_top1#0", "target_node_id": "add@switch_top1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-366", "source_node_id": "python@switch_top1.expert2#2", "target_node_id": "add@switch_top1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-367", "source_node_id": "switch_top1.expert2", "target_node_id": "add@switch_top1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-368", "source_node_id": "add@switch_top1#1", "target_node_id": "add@switch_top1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-369", "source_node_id": "python@switch_top1.expert3#2", "target_node_id": "add@switch_top1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-370", "source_node_id": "switch_top1.expert3", "target_node_id": "add@switch_top1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-371", "source_node_id": "add@switch_top1#2", "target_node_id": "add@switch_top1#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-372", "source_node_id": "python@switch_top1.expert4#2", "target_node_id": "add@switch_top1#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-373", "source_node_id": "switch_top1.expert4", "target_node_id": "add@switch_top1#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-374", "source_node_id": "add@switch_top1#3", "target_node_id": "add@switch_top1#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-375", "source_node_id": "python@switch_top1.expert5#2", "target_node_id": "add@switch_top1#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-376", "source_node_id": "switch_top1.expert5", "target_node_id": "add@switch_top1#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-377", "source_node_id": "add@switch_top1#4", "target_node_id": "add@switch_top1#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-378", "source_node_id": "python@switch_top1.expert6#2", "target_node_id": "add@switch_top1#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-379", "source_node_id": "switch_top1.expert6", "target_node_id": "add@switch_top1#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-380", "source_node_id": "add@switch_top1#5", "target_node_id": "add@switch_top1#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-381", "source_node_id": "python@switch_top1.expert7#2", "target_node_id": "add@switch_top1#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-382", "source_node_id": "switch_top1.expert7", "target_node_id": "add@switch_top1#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-383", "source_node_id": "mixtral_top2.router.noise_scale", "target_node_id": "softplus@mixtral_top2.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-384", "source_node_id": "softplus@mixtral_top2.router#0", "target_node_id": "mul@mixtral_top2.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-385", "source_node_id": "mixtral_top2.router.logits", "target_node_id": "mul@mixtral_top2.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-386", "source_node_id": "mixtral_top2.router.logits", "target_node_id": "add@mixtral_top2.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-387", "source_node_id": "mul@mixtral_top2.router#0", "target_node_id": "add@mixtral_top2.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-388", "source_node_id": "add@mixtral_top2.router#0", "target_node_id": "python@mixtral_top2.router#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-389", "source_node_id": "python@mixtral_top2.router#0", "target_node_id": "python@mixtral_top2.router#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-390", "source_node_id": "python@mixtral_top2.router#0", "target_node_id": "python@mixtral_top2.router#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-391", "source_node_id": "python@mixtral_top2.router#1", "target_node_id": "python@mixtral_top2.router#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-392", "source_node_id": "mixtral_top2.expert0.expert.gate", "target_node_id": "mixtral_top2.expert0.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-393", "source_node_id": "mixtral_top2.expert0.expert.silu", "target_node_id": "mul@mixtral_top2.expert0.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-394", "source_node_id": "mixtral_top2.expert0.expert.up", "target_node_id": "mul@mixtral_top2.expert0.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-395", "source_node_id": "mul@mixtral_top2.expert0.expert#0", "target_node_id": "mixtral_top2.expert0.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-396", "source_node_id": "python@mixtral_top2.expert0#0", "target_node_id": "python@mixtral_top2.expert0#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-397", "source_node_id": "flatten@mixtral_top2.expert0#0", "target_node_id": "python@mixtral_top2.expert0#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-398", "source_node_id": "python@mixtral_top2.expert0#1", "target_node_id": "mixtral_top2.expert0.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-399", "source_node_id": "python@mixtral_top2.expert0#1", "target_node_id": "mixtral_top2.expert0.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-400", "source_node_id": "python@mixtral_top2.expert0#1", "target_node_id": "mixtral_top2.expert0.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-401", "source_node_id": "mixtral_top2.expert0.expert.down", "target_node_id": "mul@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-402", "source_node_id": "mixtral_top2.expert0.expert", "target_node_id": "mul@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-403", "source_node_id": "flatten@mixtral_top2.expert0#0", "target_node_id": "mul@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-404", "source_node_id": "python@mixtral_top2.expert0#0", "target_node_id": "python@mixtral_top2.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-405", "source_node_id": "flatten@mixtral_top2.expert0#0", "target_node_id": "python@mixtral_top2.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-406", "source_node_id": "mul@mixtral_top2.expert0#0", "target_node_id": "python@mixtral_top2.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-407", "source_node_id": "mixtral_top2.expert1.expert.gate", "target_node_id": "mixtral_top2.expert1.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-408", "source_node_id": "mixtral_top2.expert1.expert.silu", "target_node_id": "mul@mixtral_top2.expert1.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-409", "source_node_id": "mixtral_top2.expert1.expert.up", "target_node_id": "mul@mixtral_top2.expert1.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-410", "source_node_id": "mul@mixtral_top2.expert1.expert#0", "target_node_id": "mixtral_top2.expert1.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-411", "source_node_id": "python@mixtral_top2.expert1#0", "target_node_id": "python@mixtral_top2.expert1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-412", "source_node_id": "flatten@mixtral_top2.expert1#0", "target_node_id": "python@mixtral_top2.expert1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-413", "source_node_id": "python@mixtral_top2.expert1#1", "target_node_id": "mixtral_top2.expert1.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-414", "source_node_id": "python@mixtral_top2.expert1#1", "target_node_id": "mixtral_top2.expert1.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-415", "source_node_id": "python@mixtral_top2.expert1#1", "target_node_id": "mixtral_top2.expert1.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-416", "source_node_id": "mixtral_top2.expert1.expert.down", "target_node_id": "mul@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-417", "source_node_id": "mixtral_top2.expert1.expert", "target_node_id": "mul@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-418", "source_node_id": "flatten@mixtral_top2.expert1#0", "target_node_id": "mul@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-419", "source_node_id": "python@mixtral_top2.expert1#0", "target_node_id": "python@mixtral_top2.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-420", "source_node_id": "flatten@mixtral_top2.expert1#0", "target_node_id": "python@mixtral_top2.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-421", "source_node_id": "mul@mixtral_top2.expert1#0", "target_node_id": "python@mixtral_top2.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-422", "source_node_id": "mixtral_top2.expert2.expert.gate", "target_node_id": "mixtral_top2.expert2.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-423", "source_node_id": "mixtral_top2.expert2.expert.silu", "target_node_id": "mul@mixtral_top2.expert2.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-424", "source_node_id": "mixtral_top2.expert2.expert.up", "target_node_id": "mul@mixtral_top2.expert2.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-425", "source_node_id": "mul@mixtral_top2.expert2.expert#0", "target_node_id": "mixtral_top2.expert2.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-426", "source_node_id": "python@mixtral_top2.expert2#0", "target_node_id": "python@mixtral_top2.expert2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-427", "source_node_id": "flatten@mixtral_top2.expert2#0", "target_node_id": "python@mixtral_top2.expert2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-428", "source_node_id": "python@mixtral_top2.expert2#1", "target_node_id": "mixtral_top2.expert2.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-429", "source_node_id": "python@mixtral_top2.expert2#1", "target_node_id": "mixtral_top2.expert2.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-430", "source_node_id": "python@mixtral_top2.expert2#1", "target_node_id": "mixtral_top2.expert2.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-431", "source_node_id": "mixtral_top2.expert2.expert.down", "target_node_id": "mul@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-432", "source_node_id": "mixtral_top2.expert2.expert", "target_node_id": "mul@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-433", "source_node_id": "flatten@mixtral_top2.expert2#0", "target_node_id": "mul@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-434", "source_node_id": "python@mixtral_top2.expert2#0", "target_node_id": "python@mixtral_top2.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-435", "source_node_id": "flatten@mixtral_top2.expert2#0", "target_node_id": "python@mixtral_top2.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-436", "source_node_id": "mul@mixtral_top2.expert2#0", "target_node_id": "python@mixtral_top2.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-437", "source_node_id": "mixtral_top2.expert3.expert.gate", "target_node_id": "mixtral_top2.expert3.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-438", "source_node_id": "mixtral_top2.expert3.expert.silu", "target_node_id": "mul@mixtral_top2.expert3.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-439", "source_node_id": "mixtral_top2.expert3.expert.up", "target_node_id": "mul@mixtral_top2.expert3.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-440", "source_node_id": "mul@mixtral_top2.expert3.expert#0", "target_node_id": "mixtral_top2.expert3.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-441", "source_node_id": "python@mixtral_top2.expert3#0", "target_node_id": "python@mixtral_top2.expert3#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-442", "source_node_id": "flatten@mixtral_top2.expert3#0", "target_node_id": "python@mixtral_top2.expert3#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-443", "source_node_id": "python@mixtral_top2.expert3#1", "target_node_id": "mixtral_top2.expert3.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-444", "source_node_id": "python@mixtral_top2.expert3#1", "target_node_id": "mixtral_top2.expert3.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-445", "source_node_id": "python@mixtral_top2.expert3#1", "target_node_id": "mixtral_top2.expert3.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-446", "source_node_id": "mixtral_top2.expert3.expert.down", "target_node_id": "mul@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-447", "source_node_id": "mixtral_top2.expert3.expert", "target_node_id": "mul@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-448", "source_node_id": "flatten@mixtral_top2.expert3#0", "target_node_id": "mul@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-449", "source_node_id": "python@mixtral_top2.expert3#0", "target_node_id": "python@mixtral_top2.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-450", "source_node_id": "flatten@mixtral_top2.expert3#0", "target_node_id": "python@mixtral_top2.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-451", "source_node_id": "mul@mixtral_top2.expert3#0", "target_node_id": "python@mixtral_top2.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-452", "source_node_id": "mixtral_top2.expert4.expert.gate", "target_node_id": "mixtral_top2.expert4.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-453", "source_node_id": "mixtral_top2.expert4.expert.silu", "target_node_id": "mul@mixtral_top2.expert4.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-454", "source_node_id": "mixtral_top2.expert4.expert.up", "target_node_id": "mul@mixtral_top2.expert4.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-455", "source_node_id": "mul@mixtral_top2.expert4.expert#0", "target_node_id": "mixtral_top2.expert4.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-456", "source_node_id": "python@mixtral_top2.expert4#0", "target_node_id": "python@mixtral_top2.expert4#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-457", "source_node_id": "flatten@mixtral_top2.expert4#0", "target_node_id": "python@mixtral_top2.expert4#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-458", "source_node_id": "python@mixtral_top2.expert4#1", "target_node_id": "mixtral_top2.expert4.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-459", "source_node_id": "python@mixtral_top2.expert4#1", "target_node_id": "mixtral_top2.expert4.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-460", "source_node_id": "python@mixtral_top2.expert4#1", "target_node_id": "mixtral_top2.expert4.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-461", "source_node_id": "mixtral_top2.expert4.expert.down", "target_node_id": "mul@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-462", "source_node_id": "mixtral_top2.expert4.expert", "target_node_id": "mul@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-463", "source_node_id": "flatten@mixtral_top2.expert4#0", "target_node_id": "mul@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-464", "source_node_id": "python@mixtral_top2.expert4#0", "target_node_id": "python@mixtral_top2.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-465", "source_node_id": "flatten@mixtral_top2.expert4#0", "target_node_id": "python@mixtral_top2.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-466", "source_node_id": "mul@mixtral_top2.expert4#0", "target_node_id": "python@mixtral_top2.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-467", "source_node_id": "mixtral_top2.expert5.expert.gate", "target_node_id": "mixtral_top2.expert5.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-468", "source_node_id": "mixtral_top2.expert5.expert.silu", "target_node_id": "mul@mixtral_top2.expert5.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-469", "source_node_id": "mixtral_top2.expert5.expert.up", "target_node_id": "mul@mixtral_top2.expert5.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-470", "source_node_id": "mul@mixtral_top2.expert5.expert#0", "target_node_id": "mixtral_top2.expert5.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-471", "source_node_id": "python@mixtral_top2.expert5#0", "target_node_id": "python@mixtral_top2.expert5#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-472", "source_node_id": "flatten@mixtral_top2.expert5#0", "target_node_id": "python@mixtral_top2.expert5#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-473", "source_node_id": "python@mixtral_top2.expert5#1", "target_node_id": "mixtral_top2.expert5.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-474", "source_node_id": "python@mixtral_top2.expert5#1", "target_node_id": "mixtral_top2.expert5.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-475", "source_node_id": "python@mixtral_top2.expert5#1", "target_node_id": "mixtral_top2.expert5.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-476", "source_node_id": "mixtral_top2.expert5.expert.down", "target_node_id": "mul@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-477", "source_node_id": "mixtral_top2.expert5.expert", "target_node_id": "mul@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-478", "source_node_id": "flatten@mixtral_top2.expert5#0", "target_node_id": "mul@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-479", "source_node_id": "python@mixtral_top2.expert5#0", "target_node_id": "python@mixtral_top2.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-480", "source_node_id": "flatten@mixtral_top2.expert5#0", "target_node_id": "python@mixtral_top2.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-481", "source_node_id": "mul@mixtral_top2.expert5#0", "target_node_id": "python@mixtral_top2.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-482", "source_node_id": "mixtral_top2.expert6.expert.gate", "target_node_id": "mixtral_top2.expert6.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-483", "source_node_id": "mixtral_top2.expert6.expert.silu", "target_node_id": "mul@mixtral_top2.expert6.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-484", "source_node_id": "mixtral_top2.expert6.expert.up", "target_node_id": "mul@mixtral_top2.expert6.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-485", "source_node_id": "mul@mixtral_top2.expert6.expert#0", "target_node_id": "mixtral_top2.expert6.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-486", "source_node_id": "python@mixtral_top2.expert6#0", "target_node_id": "python@mixtral_top2.expert6#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-487", "source_node_id": "flatten@mixtral_top2.expert6#0", "target_node_id": "python@mixtral_top2.expert6#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-488", "source_node_id": "python@mixtral_top2.expert6#1", "target_node_id": "mixtral_top2.expert6.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-489", "source_node_id": "python@mixtral_top2.expert6#1", "target_node_id": "mixtral_top2.expert6.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-490", "source_node_id": "python@mixtral_top2.expert6#1", "target_node_id": "mixtral_top2.expert6.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-491", "source_node_id": "mixtral_top2.expert6.expert.down", "target_node_id": "mul@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-492", "source_node_id": "mixtral_top2.expert6.expert", "target_node_id": "mul@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-493", "source_node_id": "flatten@mixtral_top2.expert6#0", "target_node_id": "mul@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-494", "source_node_id": "python@mixtral_top2.expert6#0", "target_node_id": "python@mixtral_top2.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-495", "source_node_id": "flatten@mixtral_top2.expert6#0", "target_node_id": "python@mixtral_top2.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-496", "source_node_id": "mul@mixtral_top2.expert6#0", "target_node_id": "python@mixtral_top2.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-497", "source_node_id": "mixtral_top2.expert7.expert.gate", "target_node_id": "mixtral_top2.expert7.expert.silu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-498", "source_node_id": "mixtral_top2.expert7.expert.silu", "target_node_id": "mul@mixtral_top2.expert7.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-499", "source_node_id": "mixtral_top2.expert7.expert.up", "target_node_id": "mul@mixtral_top2.expert7.expert#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-500", "source_node_id": "mul@mixtral_top2.expert7.expert#0", "target_node_id": "mixtral_top2.expert7.expert.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-501", "source_node_id": "python@mixtral_top2.expert7#0", "target_node_id": "python@mixtral_top2.expert7#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-502", "source_node_id": "flatten@mixtral_top2.expert7#0", "target_node_id": "python@mixtral_top2.expert7#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-503", "source_node_id": "python@mixtral_top2.expert7#1", "target_node_id": "mixtral_top2.expert7.expert.gate", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-504", "source_node_id": "python@mixtral_top2.expert7#1", "target_node_id": "mixtral_top2.expert7.expert.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-505", "source_node_id": "python@mixtral_top2.expert7#1", "target_node_id": "mixtral_top2.expert7.expert", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-506", "source_node_id": "mixtral_top2.expert7.expert.down", "target_node_id": "mul@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-507", "source_node_id": "mixtral_top2.expert7.expert", "target_node_id": "mul@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-508", "source_node_id": "flatten@mixtral_top2.expert7#0", "target_node_id": "mul@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-509", "source_node_id": "python@mixtral_top2.expert7#0", "target_node_id": "python@mixtral_top2.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-510", "source_node_id": "flatten@mixtral_top2.expert7#0", "target_node_id": "python@mixtral_top2.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-511", "source_node_id": "mul@mixtral_top2.expert7#0", "target_node_id": "python@mixtral_top2.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-512", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-513", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-514", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-515", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-516", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-517", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-518", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-519", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-520", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-521", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-522", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-523", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-524", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-525", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-526", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-527", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-528", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-529", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-530", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-531", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-532", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-533", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-534", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-535", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-536", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-537", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-538", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-539", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-540", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-541", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-542", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-543", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-544", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-545", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-546", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-547", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-548", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "flatten@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-549", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "mul@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-550", "source_node_id": "python@mixtral_top2.router#2", "target_node_id": "python@mixtral_top2.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-551", "source_node_id": "mixtral_top2.router", "target_node_id": "mixtral_top2.expert7", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-552", "source_node_id": "python@mixtral_top2.expert0#2", "target_node_id": "add@mixtral_top2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-553", "source_node_id": "mixtral_top2.expert0", "target_node_id": "add@mixtral_top2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-554", "source_node_id": "python@mixtral_top2.expert1#2", "target_node_id": "add@mixtral_top2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-555", "source_node_id": "mixtral_top2.expert1", "target_node_id": "add@mixtral_top2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-556", "source_node_id": "add@mixtral_top2#0", "target_node_id": "add@mixtral_top2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-557", "source_node_id": "python@mixtral_top2.expert2#2", "target_node_id": "add@mixtral_top2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-558", "source_node_id": "mixtral_top2.expert2", "target_node_id": "add@mixtral_top2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-559", "source_node_id": "add@mixtral_top2#1", "target_node_id": "add@mixtral_top2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-560", "source_node_id": "python@mixtral_top2.expert3#2", "target_node_id": "add@mixtral_top2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-561", "source_node_id": "mixtral_top2.expert3", "target_node_id": "add@mixtral_top2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-562", "source_node_id": "add@mixtral_top2#2", "target_node_id": "add@mixtral_top2#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-563", "source_node_id": "python@mixtral_top2.expert4#2", "target_node_id": "add@mixtral_top2#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-564", "source_node_id": "mixtral_top2.expert4", "target_node_id": "add@mixtral_top2#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-565", "source_node_id": "add@mixtral_top2#3", "target_node_id": "add@mixtral_top2#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-566", "source_node_id": "python@mixtral_top2.expert5#2", "target_node_id": "add@mixtral_top2#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-567", "source_node_id": "mixtral_top2.expert5", "target_node_id": "add@mixtral_top2#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-568", "source_node_id": "add@mixtral_top2#4", "target_node_id": "add@mixtral_top2#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-569", "source_node_id": "python@mixtral_top2.expert6#2", "target_node_id": "add@mixtral_top2#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-570", "source_node_id": "mixtral_top2.expert6", "target_node_id": "add@mixtral_top2#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-571", "source_node_id": "add@mixtral_top2#5", "target_node_id": "add@mixtral_top2#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-572", "source_node_id": "python@mixtral_top2.expert7#2", "target_node_id": "add@mixtral_top2#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-573", "source_node_id": "mixtral_top2.expert7", "target_node_id": "add@mixtral_top2#6", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-574", "source_node_id": "input", "target_node_id": "noisy_topk.router.logits", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-575", "source_node_id": "input", "target_node_id": "noisy_topk.router.noise_scale", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-576", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-577", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-578", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-579", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-580", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-581", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-582", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-583", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-584", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-585", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-586", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-587", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-588", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-589", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-590", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-591", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-592", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-593", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-594", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-595", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-596", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-597", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-598", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-599", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-600", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-601", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-602", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-603", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-604", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-605", "source_node_id": "input", "target_node_id": "flatten@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-606", "source_node_id": "input", "target_node_id": "mul@noisy_topk.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-607", "source_node_id": "input", "target_node_id": "python@noisy_topk.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-608", "source_node_id": "input", "target_node_id": "noisy_topk", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-609", "source_node_id": "input", "target_node_id": "switch_top1.router.logits", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-610", "source_node_id": "input", "target_node_id": "switch_top1.router.noise_scale", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-611", "source_node_id": "input", "target_node_id": "python@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-612", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-613", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-614", "source_node_id": "input", "target_node_id": "python@switch_top1.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-615", "source_node_id": "input", "target_node_id": "python@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-616", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-617", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-618", "source_node_id": "input", "target_node_id": "python@switch_top1.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-619", "source_node_id": "input", "target_node_id": "python@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-620", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-621", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-622", "source_node_id": "input", "target_node_id": "python@switch_top1.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-623", "source_node_id": "input", "target_node_id": "python@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-624", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-625", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-626", "source_node_id": "input", "target_node_id": "python@switch_top1.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-627", "source_node_id": "input", "target_node_id": "python@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-628", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-629", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-630", "source_node_id": "input", "target_node_id": "python@switch_top1.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-631", "source_node_id": "input", "target_node_id": "python@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-632", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-633", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-634", "source_node_id": "input", "target_node_id": "python@switch_top1.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-635", "source_node_id": "input", "target_node_id": "python@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-636", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-637", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-638", "source_node_id": "input", "target_node_id": "python@switch_top1.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-639", "source_node_id": "input", "target_node_id": "python@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-640", "source_node_id": "input", "target_node_id": "flatten@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-641", "source_node_id": "input", "target_node_id": "mul@switch_top1.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-642", "source_node_id": "input", "target_node_id": "python@switch_top1.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-643", "source_node_id": "input", "target_node_id": "switch_top1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-644", "source_node_id": "input", "target_node_id": "mixtral_top2.router.logits", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-645", "source_node_id": "input", "target_node_id": "mixtral_top2.router.noise_scale", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-646", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-647", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-648", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-649", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert0#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-650", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-651", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-652", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-653", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert1#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-654", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-655", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-656", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-657", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert2#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-658", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-659", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-660", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert3#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-661", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert3#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-662", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-663", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-664", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert4#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-665", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert4#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-666", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-667", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-668", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert5#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-669", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert5#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-670", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-671", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-672", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert6#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-673", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert6#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-674", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-675", "source_node_id": "input", "target_node_id": "flatten@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-676", "source_node_id": "input", "target_node_id": "mul@mixtral_top2.expert7#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-677", "source_node_id": "input", "target_node_id": "python@mixtral_top2.expert7#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-678", "source_node_id": "input", "target_node_id": "mixtral_top2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-679", "source_node_id": "add@noisy_topk#6", "target_node_id": "stack@SparseRoutingComparison#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-680", "source_node_id": "noisy_topk", "target_node_id": "stack@SparseRoutingComparison#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-681", "source_node_id": "add@switch_top1#6", "target_node_id": "stack@SparseRoutingComparison#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-682", "source_node_id": "switch_top1", "target_node_id": "stack@SparseRoutingComparison#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-683", "source_node_id": "add@mixtral_top2#6", "target_node_id": "stack@SparseRoutingComparison#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-684", "source_node_id": "mixtral_top2", "target_node_id": "stack@SparseRoutingComparison#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-685", "source_node_id": "stack@SparseRoutingComparison#0", "target_node_id": "output", "source_port": "y", "target_port": "x", "metadata": {} } ], "inputs": [], "modules": [], "operations": [], "metadata": {} }, "view": { "graph_id": "snapshot", "node_positions": { "input": { "x": 0, "y": 160 }, "noisy_topk": { "x": 340, "y": 160 }, "switch_top1": { "x": 680, "y": 160 }, "mixtral_top2": { "x": 1020, "y": 160 }, "stack@SparseRoutingComparison#0": { "x": 1360, "y": 160 }, "output": { "x": 1700, "y": 160 }, "noisy_topk.router": { "x": 24, "y": 214 }, "noisy_topk.expert0": { "x": 24, "y": 364 }, "noisy_topk.expert1": { "x": 24, "y": 514 }, "add@noisy_topk#0": { "x": 24, "y": 664 }, "noisy_topk.expert2": { "x": 24, "y": 814 }, "add@noisy_topk#1": { "x": 24, "y": 964 }, "noisy_topk.expert3": { "x": 24, "y": 1114 }, "add@noisy_topk#2": { "x": 24, "y": 1264 }, "noisy_topk.expert4": { "x": 24, "y": 1414 }, "add@noisy_topk#3": { "x": 24, "y": 1564 }, "noisy_topk.expert5": { "x": 24, "y": 1714 }, "add@noisy_topk#4": { "x": 24, "y": 1864 }, "noisy_topk.expert6": { "x": 24, "y": 2014 }, "add@noisy_topk#5": { "x": 24, "y": 2164 }, "noisy_topk.expert7": { "x": 24, "y": 2314 }, "add@noisy_topk#6": { "x": 24, "y": 2464 }, "noisy_topk.router.logits": { "x": 24, "y": 214 }, "noisy_topk.router.noise_scale": { "x": 24, "y": 364 }, "softplus@noisy_topk.router#0": { "x": 24, "y": 514 }, "mul@noisy_topk.router#0": { "x": 24, "y": 664 }, "add@noisy_topk.router#0": { "x": 24, "y": 814 }, "python@noisy_topk.router#0": { "x": 24, "y": 964 }, "python@noisy_topk.router#1": { "x": 24, "y": 1114 }, "python@noisy_topk.router#2": { "x": 24, "y": 1264 }, "flatten@noisy_topk.expert0#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert0#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert0#1": { "x": 24, "y": 514 }, "noisy_topk.expert0.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert0#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert0#2": { "x": 24, "y": 964 }, "noisy_topk.expert0.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert0.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert0.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert1#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert1#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert1#1": { "x": 24, "y": 514 }, "noisy_topk.expert1.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert1#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert1#2": { "x": 24, "y": 964 }, "noisy_topk.expert1.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert1.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert1.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert2#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert2#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert2#1": { "x": 24, "y": 514 }, "noisy_topk.expert2.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert2#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert2#2": { "x": 24, "y": 964 }, "noisy_topk.expert2.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert2.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert2.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert3#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert3#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert3#1": { "x": 24, "y": 514 }, "noisy_topk.expert3.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert3#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert3#2": { "x": 24, "y": 964 }, "noisy_topk.expert3.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert3.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert3.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert4#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert4#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert4#1": { "x": 24, "y": 514 }, "noisy_topk.expert4.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert4#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert4#2": { "x": 24, "y": 964 }, "noisy_topk.expert4.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert4.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert4.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert5#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert5#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert5#1": { "x": 24, "y": 514 }, "noisy_topk.expert5.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert5#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert5#2": { "x": 24, "y": 964 }, "noisy_topk.expert5.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert5.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert5.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert6#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert6#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert6#1": { "x": 24, "y": 514 }, "noisy_topk.expert6.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert6#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert6#2": { "x": 24, "y": 964 }, "noisy_topk.expert6.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert6.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert6.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert7#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert7#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert7#1": { "x": 24, "y": 514 }, "noisy_topk.expert7.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert7#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert7#2": { "x": 24, "y": 964 }, "noisy_topk.expert7.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert7.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert7.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert7.expert.down": { "x": 24, "y": 814 }, "switch_top1.router": { "x": 24, "y": 214 }, "switch_top1.expert0": { "x": 24, "y": 364 }, "switch_top1.expert1": { "x": 24, "y": 514 }, "add@switch_top1#0": { "x": 24, "y": 664 }, "switch_top1.expert2": { "x": 24, "y": 814 }, "add@switch_top1#1": { "x": 24, "y": 964 }, "switch_top1.expert3": { "x": 24, "y": 1114 }, "add@switch_top1#2": { "x": 24, "y": 1264 }, "switch_top1.expert4": { "x": 24, "y": 1414 }, "add@switch_top1#3": { "x": 24, "y": 1564 }, "switch_top1.expert5": { "x": 24, "y": 1714 }, "add@switch_top1#4": { "x": 24, "y": 1864 }, "switch_top1.expert6": { "x": 24, "y": 2014 }, "add@switch_top1#5": { "x": 24, "y": 2164 }, "switch_top1.expert7": { "x": 24, "y": 2314 }, "add@switch_top1#6": { "x": 24, "y": 2464 }, "switch_top1.router.logits": { "x": 24, "y": 214 }, "switch_top1.router.noise_scale": { "x": 24, "y": 364 }, "softplus@switch_top1.router#0": { "x": 24, "y": 514 }, "mul@switch_top1.router#0": { "x": 24, "y": 664 }, "add@switch_top1.router#0": { "x": 24, "y": 814 }, "python@switch_top1.router#0": { "x": 24, "y": 964 }, "python@switch_top1.router#1": { "x": 24, "y": 1114 }, "python@switch_top1.router#2": { "x": 24, "y": 1264 }, "flatten@switch_top1.expert0#0": { "x": 24, "y": 214 }, "python@switch_top1.expert0#0": { "x": 24, "y": 364 }, "python@switch_top1.expert0#1": { "x": 24, "y": 514 }, "switch_top1.expert0.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert0#0": { "x": 24, "y": 814 }, "python@switch_top1.expert0#2": { "x": 24, "y": 964 }, "switch_top1.expert0.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert0.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert0.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert1#0": { "x": 24, "y": 214 }, "python@switch_top1.expert1#0": { "x": 24, "y": 364 }, "python@switch_top1.expert1#1": { "x": 24, "y": 514 }, "switch_top1.expert1.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert1#0": { "x": 24, "y": 814 }, "python@switch_top1.expert1#2": { "x": 24, "y": 964 }, "switch_top1.expert1.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert1.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert1.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert2#0": { "x": 24, "y": 214 }, "python@switch_top1.expert2#0": { "x": 24, "y": 364 }, "python@switch_top1.expert2#1": { "x": 24, "y": 514 }, "switch_top1.expert2.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert2#0": { "x": 24, "y": 814 }, "python@switch_top1.expert2#2": { "x": 24, "y": 964 }, "switch_top1.expert2.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert2.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert2.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert3#0": { "x": 24, "y": 214 }, "python@switch_top1.expert3#0": { "x": 24, "y": 364 }, "python@switch_top1.expert3#1": { "x": 24, "y": 514 }, "switch_top1.expert3.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert3#0": { "x": 24, "y": 814 }, "python@switch_top1.expert3#2": { "x": 24, "y": 964 }, "switch_top1.expert3.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert3.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert3.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert4#0": { "x": 24, "y": 214 }, "python@switch_top1.expert4#0": { "x": 24, "y": 364 }, "python@switch_top1.expert4#1": { "x": 24, "y": 514 }, "switch_top1.expert4.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert4#0": { "x": 24, "y": 814 }, "python@switch_top1.expert4#2": { "x": 24, "y": 964 }, "switch_top1.expert4.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert4.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert4.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert5#0": { "x": 24, "y": 214 }, "python@switch_top1.expert5#0": { "x": 24, "y": 364 }, "python@switch_top1.expert5#1": { "x": 24, "y": 514 }, "switch_top1.expert5.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert5#0": { "x": 24, "y": 814 }, "python@switch_top1.expert5#2": { "x": 24, "y": 964 }, "switch_top1.expert5.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert5.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert5.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert6#0": { "x": 24, "y": 214 }, "python@switch_top1.expert6#0": { "x": 24, "y": 364 }, "python@switch_top1.expert6#1": { "x": 24, "y": 514 }, "switch_top1.expert6.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert6#0": { "x": 24, "y": 814 }, "python@switch_top1.expert6#2": { "x": 24, "y": 964 }, "switch_top1.expert6.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert6.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert6.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert7#0": { "x": 24, "y": 214 }, "python@switch_top1.expert7#0": { "x": 24, "y": 364 }, "python@switch_top1.expert7#1": { "x": 24, "y": 514 }, "switch_top1.expert7.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert7#0": { "x": 24, "y": 814 }, "python@switch_top1.expert7#2": { "x": 24, "y": 964 }, "switch_top1.expert7.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert7.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert7.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert7.expert.down": { "x": 24, "y": 814 }, "mixtral_top2.router": { "x": 24, "y": 214 }, "mixtral_top2.expert0": { "x": 24, "y": 364 }, "mixtral_top2.expert1": { "x": 24, "y": 514 }, "add@mixtral_top2#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2": { "x": 24, "y": 814 }, "add@mixtral_top2#1": { "x": 24, "y": 964 }, "mixtral_top2.expert3": { "x": 24, "y": 1114 }, "add@mixtral_top2#2": { "x": 24, "y": 1264 }, "mixtral_top2.expert4": { "x": 24, "y": 1414 }, "add@mixtral_top2#3": { "x": 24, "y": 1564 }, "mixtral_top2.expert5": { "x": 24, "y": 1714 }, "add@mixtral_top2#4": { "x": 24, "y": 1864 }, "mixtral_top2.expert6": { "x": 24, "y": 2014 }, "add@mixtral_top2#5": { "x": 24, "y": 2164 }, "mixtral_top2.expert7": { "x": 24, "y": 2314 }, "add@mixtral_top2#6": { "x": 24, "y": 2464 }, "mixtral_top2.router.logits": { "x": 24, "y": 214 }, "mixtral_top2.router.noise_scale": { "x": 24, "y": 364 }, "softplus@mixtral_top2.router#0": { "x": 24, "y": 514 }, "mul@mixtral_top2.router#0": { "x": 24, "y": 664 }, "add@mixtral_top2.router#0": { "x": 24, "y": 814 }, "python@mixtral_top2.router#0": { "x": 24, "y": 964 }, "python@mixtral_top2.router#1": { "x": 24, "y": 1114 }, "python@mixtral_top2.router#2": { "x": 24, "y": 1264 }, "flatten@mixtral_top2.expert0#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert0#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert0#1": { "x": 24, "y": 514 }, "mixtral_top2.expert0.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert0#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert0#2": { "x": 24, "y": 964 }, "mixtral_top2.expert0.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert0.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert0.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert1#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert1#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert1#1": { "x": 24, "y": 514 }, "mixtral_top2.expert1.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert1#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert1#2": { "x": 24, "y": 964 }, "mixtral_top2.expert1.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert1.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert1.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert2#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert2#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert2#1": { "x": 24, "y": 514 }, "mixtral_top2.expert2.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert2#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert2#2": { "x": 24, "y": 964 }, "mixtral_top2.expert2.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert2.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert2.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert3#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert3#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert3#1": { "x": 24, "y": 514 }, "mixtral_top2.expert3.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert3#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert3#2": { "x": 24, "y": 964 }, "mixtral_top2.expert3.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert3.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert3.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert4#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert4#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert4#1": { "x": 24, "y": 514 }, "mixtral_top2.expert4.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert4#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert4#2": { "x": 24, "y": 964 }, "mixtral_top2.expert4.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert4.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert4.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert5#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert5#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert5#1": { "x": 24, "y": 514 }, "mixtral_top2.expert5.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert5#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert5#2": { "x": 24, "y": 964 }, "mixtral_top2.expert5.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert5.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert5.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert6#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert6#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert6#1": { "x": 24, "y": 514 }, "mixtral_top2.expert6.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert6#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert6#2": { "x": 24, "y": 964 }, "mixtral_top2.expert6.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert6.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert6.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert7#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert7#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert7#1": { "x": 24, "y": 514 }, "mixtral_top2.expert7.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert7#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert7#2": { "x": 24, "y": 964 }, "mixtral_top2.expert7.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert7.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert7.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert7.expert.down": { "x": 24, "y": 814 } }, "collapsed_modules": [ "noisy_topk", "switch_top1", "mixtral_top2", "noisy_topk.router", "noisy_topk.expert0", "noisy_topk.expert1", "noisy_topk.expert2", "noisy_topk.expert3", "noisy_topk.expert4", "noisy_topk.expert5", "noisy_topk.expert6", "noisy_topk.expert7", "noisy_topk.expert0.expert", "noisy_topk.expert1.expert", "noisy_topk.expert2.expert", "noisy_topk.expert3.expert", "noisy_topk.expert4.expert", "noisy_topk.expert5.expert", "noisy_topk.expert6.expert", "noisy_topk.expert7.expert", "switch_top1.router", "switch_top1.expert0", "switch_top1.expert1", "switch_top1.expert2", "switch_top1.expert3", "switch_top1.expert4", "switch_top1.expert5", "switch_top1.expert6", "switch_top1.expert7", "switch_top1.expert0.expert", "switch_top1.expert1.expert", "switch_top1.expert2.expert", "switch_top1.expert3.expert", "switch_top1.expert4.expert", "switch_top1.expert5.expert", "switch_top1.expert6.expert", "switch_top1.expert7.expert", "mixtral_top2.router", "mixtral_top2.expert0", "mixtral_top2.expert1", "mixtral_top2.expert2", "mixtral_top2.expert3", "mixtral_top2.expert4", "mixtral_top2.expert5", "mixtral_top2.expert6", "mixtral_top2.expert7", "mixtral_top2.expert0.expert", "mixtral_top2.expert1.expert", "mixtral_top2.expert2.expert", "mixtral_top2.expert3.expert", "mixtral_top2.expert4.expert", "mixtral_top2.expert5.expert", "mixtral_top2.expert6.expert", "mixtral_top2.expert7.expert" ], "selection": [], "layout": {} } }, "snapshotId": "snapshot-82f8c493b242ec59", "collaboration": { "walkthrough": { "title": "Let a router choose the experts", "steps": [ { "id": "route", "title": "Score and select two experts", "note": "The router scores eight experts, selects the two largest probabilities and renormalizes those selected weights. Each token can choose a different pair. The recorded diagram shows actual routes for synthetic inputs.", "view": { "camera": { "x": 291, "y": -4, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "noisy_topk": { "x": 340, "y": 160 }, "switch_top1": { "x": 680, "y": 160 }, "mixtral_top2": { "x": 1020, "y": 160 }, "stack@SparseRoutingComparison#0": { "x": 1360, "y": 160 }, "output": { "x": 1700, "y": 160 }, "noisy_topk.router": { "x": 24, "y": 214 }, "noisy_topk.expert0": { "x": 24, "y": 364 }, "noisy_topk.expert1": { "x": 24, "y": 514 }, "add@noisy_topk#0": { "x": 24, "y": 664 }, "noisy_topk.expert2": { "x": 24, "y": 814 }, "add@noisy_topk#1": { "x": 24, "y": 964 }, "noisy_topk.expert3": { "x": 24, "y": 1114 }, "add@noisy_topk#2": { "x": 24, "y": 1264 }, "noisy_topk.expert4": { "x": 24, "y": 1414 }, "add@noisy_topk#3": { "x": 24, "y": 1564 }, "noisy_topk.expert5": { "x": 24, "y": 1714 }, "add@noisy_topk#4": { "x": 24, "y": 1864 }, "noisy_topk.expert6": { "x": 24, "y": 2014 }, "add@noisy_topk#5": { "x": 24, "y": 2164 }, "noisy_topk.expert7": { "x": 24, "y": 2314 }, "add@noisy_topk#6": { "x": 24, "y": 2464 }, "noisy_topk.router.logits": { "x": 24, "y": 214 }, "noisy_topk.router.noise_scale": { "x": 24, "y": 364 }, "softplus@noisy_topk.router#0": { "x": 24, "y": 514 }, "mul@noisy_topk.router#0": { "x": 24, "y": 664 }, "add@noisy_topk.router#0": { "x": 24, "y": 814 }, "python@noisy_topk.router#0": { "x": 24, "y": 964 }, "python@noisy_topk.router#1": { "x": 24, "y": 1114 }, "python@noisy_topk.router#2": { "x": 24, "y": 1264 }, "flatten@noisy_topk.expert0#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert0#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert0#1": { "x": 24, "y": 514 }, "noisy_topk.expert0.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert0#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert0#2": { "x": 24, "y": 964 }, "noisy_topk.expert0.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert0.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert0.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert1#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert1#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert1#1": { "x": 24, "y": 514 }, "noisy_topk.expert1.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert1#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert1#2": { "x": 24, "y": 964 }, "noisy_topk.expert1.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert1.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert1.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert2#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert2#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert2#1": { "x": 24, "y": 514 }, "noisy_topk.expert2.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert2#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert2#2": { "x": 24, "y": 964 }, "noisy_topk.expert2.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert2.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert2.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert3#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert3#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert3#1": { "x": 24, "y": 514 }, "noisy_topk.expert3.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert3#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert3#2": { "x": 24, "y": 964 }, "noisy_topk.expert3.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert3.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert3.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert4#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert4#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert4#1": { "x": 24, "y": 514 }, "noisy_topk.expert4.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert4#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert4#2": { "x": 24, "y": 964 }, "noisy_topk.expert4.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert4.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert4.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert5#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert5#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert5#1": { "x": 24, "y": 514 }, "noisy_topk.expert5.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert5#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert5#2": { "x": 24, "y": 964 }, "noisy_topk.expert5.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert5.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert5.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert6#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert6#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert6#1": { "x": 24, "y": 514 }, "noisy_topk.expert6.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert6#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert6#2": { "x": 24, "y": 964 }, "noisy_topk.expert6.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert6.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert6.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert7#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert7#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert7#1": { "x": 24, "y": 514 }, "noisy_topk.expert7.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert7#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert7#2": { "x": 24, "y": 964 }, "noisy_topk.expert7.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert7.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert7.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert7.expert.down": { "x": 24, "y": 814 }, "switch_top1.router": { "x": 24, "y": 214 }, "switch_top1.expert0": { "x": 24, "y": 364 }, "switch_top1.expert1": { "x": 24, "y": 514 }, "add@switch_top1#0": { "x": 24, "y": 664 }, "switch_top1.expert2": { "x": 24, "y": 814 }, "add@switch_top1#1": { "x": 24, "y": 964 }, "switch_top1.expert3": { "x": 24, "y": 1114 }, "add@switch_top1#2": { "x": 24, "y": 1264 }, "switch_top1.expert4": { "x": 24, "y": 1414 }, "add@switch_top1#3": { "x": 24, "y": 1564 }, "switch_top1.expert5": { "x": 24, "y": 1714 }, "add@switch_top1#4": { "x": 24, "y": 1864 }, "switch_top1.expert6": { "x": 24, "y": 2014 }, "add@switch_top1#5": { "x": 24, "y": 2164 }, "switch_top1.expert7": { "x": 24, "y": 2314 }, "add@switch_top1#6": { "x": 24, "y": 2464 }, "switch_top1.router.logits": { "x": 24, "y": 214 }, "switch_top1.router.noise_scale": { "x": 24, "y": 364 }, "softplus@switch_top1.router#0": { "x": 24, "y": 514 }, "mul@switch_top1.router#0": { "x": 24, "y": 664 }, "add@switch_top1.router#0": { "x": 24, "y": 814 }, "python@switch_top1.router#0": { "x": 24, "y": 964 }, "python@switch_top1.router#1": { "x": 24, "y": 1114 }, "python@switch_top1.router#2": { "x": 24, "y": 1264 }, "flatten@switch_top1.expert0#0": { "x": 24, "y": 214 }, "python@switch_top1.expert0#0": { "x": 24, "y": 364 }, "python@switch_top1.expert0#1": { "x": 24, "y": 514 }, "switch_top1.expert0.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert0#0": { "x": 24, "y": 814 }, "python@switch_top1.expert0#2": { "x": 24, "y": 964 }, "switch_top1.expert0.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert0.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert0.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert1#0": { "x": 24, "y": 214 }, "python@switch_top1.expert1#0": { "x": 24, "y": 364 }, "python@switch_top1.expert1#1": { "x": 24, "y": 514 }, "switch_top1.expert1.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert1#0": { "x": 24, "y": 814 }, "python@switch_top1.expert1#2": { "x": 24, "y": 964 }, "switch_top1.expert1.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert1.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert1.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert2#0": { "x": 24, "y": 214 }, "python@switch_top1.expert2#0": { "x": 24, "y": 364 }, "python@switch_top1.expert2#1": { "x": 24, "y": 514 }, "switch_top1.expert2.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert2#0": { "x": 24, "y": 814 }, "python@switch_top1.expert2#2": { "x": 24, "y": 964 }, "switch_top1.expert2.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert2.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert2.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert3#0": { "x": 24, "y": 214 }, "python@switch_top1.expert3#0": { "x": 24, "y": 364 }, "python@switch_top1.expert3#1": { "x": 24, "y": 514 }, "switch_top1.expert3.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert3#0": { "x": 24, "y": 814 }, "python@switch_top1.expert3#2": { "x": 24, "y": 964 }, "switch_top1.expert3.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert3.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert3.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert4#0": { "x": 24, "y": 214 }, "python@switch_top1.expert4#0": { "x": 24, "y": 364 }, "python@switch_top1.expert4#1": { "x": 24, "y": 514 }, "switch_top1.expert4.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert4#0": { "x": 24, "y": 814 }, "python@switch_top1.expert4#2": { "x": 24, "y": 964 }, "switch_top1.expert4.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert4.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert4.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert5#0": { "x": 24, "y": 214 }, "python@switch_top1.expert5#0": { "x": 24, "y": 364 }, "python@switch_top1.expert5#1": { "x": 24, "y": 514 }, "switch_top1.expert5.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert5#0": { "x": 24, "y": 814 }, "python@switch_top1.expert5#2": { "x": 24, "y": 964 }, "switch_top1.expert5.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert5.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert5.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert6#0": { "x": 24, "y": 214 }, "python@switch_top1.expert6#0": { "x": 24, "y": 364 }, "python@switch_top1.expert6#1": { "x": 24, "y": 514 }, "switch_top1.expert6.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert6#0": { "x": 24, "y": 814 }, "python@switch_top1.expert6#2": { "x": 24, "y": 964 }, "switch_top1.expert6.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert6.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert6.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert7#0": { "x": 24, "y": 214 }, "python@switch_top1.expert7#0": { "x": 24, "y": 364 }, "python@switch_top1.expert7#1": { "x": 24, "y": 514 }, "switch_top1.expert7.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert7#0": { "x": 24, "y": 814 }, "python@switch_top1.expert7#2": { "x": 24, "y": 964 }, "switch_top1.expert7.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert7.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert7.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert7.expert.down": { "x": 24, "y": 814 }, "mixtral_top2.router": { "x": 24, "y": 214 }, "mixtral_top2.expert0": { "x": 24, "y": 364 }, "mixtral_top2.expert1": { "x": 24, "y": 514 }, "add@mixtral_top2#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2": { "x": 24, "y": 814 }, "add@mixtral_top2#1": { "x": 24, "y": 964 }, "mixtral_top2.expert3": { "x": 24, "y": 1114 }, "add@mixtral_top2#2": { "x": 24, "y": 1264 }, "mixtral_top2.expert4": { "x": 24, "y": 1414 }, "add@mixtral_top2#3": { "x": 24, "y": 1564 }, "mixtral_top2.expert5": { "x": 24, "y": 1714 }, "add@mixtral_top2#4": { "x": 24, "y": 1864 }, "mixtral_top2.expert6": { "x": 24, "y": 2014 }, "add@mixtral_top2#5": { "x": 24, "y": 2164 }, "mixtral_top2.expert7": { "x": 24, "y": 2314 }, "add@mixtral_top2#6": { "x": 24, "y": 2464 }, "mixtral_top2.router.logits": { "x": 24, "y": 214 }, "mixtral_top2.router.noise_scale": { "x": 24, "y": 364 }, "softplus@mixtral_top2.router#0": { "x": 24, "y": 514 }, "mul@mixtral_top2.router#0": { "x": 24, "y": 664 }, "add@mixtral_top2.router#0": { "x": 24, "y": 814 }, "python@mixtral_top2.router#0": { "x": 24, "y": 964 }, "python@mixtral_top2.router#1": { "x": 24, "y": 1114 }, "python@mixtral_top2.router#2": { "x": 24, "y": 1264 }, "flatten@mixtral_top2.expert0#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert0#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert0#1": { "x": 24, "y": 514 }, "mixtral_top2.expert0.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert0#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert0#2": { "x": 24, "y": 964 }, "mixtral_top2.expert0.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert0.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert0.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert1#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert1#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert1#1": { "x": 24, "y": 514 }, "mixtral_top2.expert1.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert1#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert1#2": { "x": 24, "y": 964 }, "mixtral_top2.expert1.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert1.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert1.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert2#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert2#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert2#1": { "x": 24, "y": 514 }, "mixtral_top2.expert2.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert2#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert2#2": { "x": 24, "y": 964 }, "mixtral_top2.expert2.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert2.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert2.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert3#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert3#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert3#1": { "x": 24, "y": 514 }, "mixtral_top2.expert3.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert3#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert3#2": { "x": 24, "y": 964 }, "mixtral_top2.expert3.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert3.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert3.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert4#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert4#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert4#1": { "x": 24, "y": 514 }, "mixtral_top2.expert4.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert4#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert4#2": { "x": 24, "y": 964 }, "mixtral_top2.expert4.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert4.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert4.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert5#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert5#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert5#1": { "x": 24, "y": 514 }, "mixtral_top2.expert5.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert5#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert5#2": { "x": 24, "y": 964 }, "mixtral_top2.expert5.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert5.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert5.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert6#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert6#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert6#1": { "x": 24, "y": 514 }, "mixtral_top2.expert6.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert6#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert6#2": { "x": 24, "y": 964 }, "mixtral_top2.expert6.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert6.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert6.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert7#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert7#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert7#1": { "x": 24, "y": 514 }, "mixtral_top2.expert7.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert7#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert7#2": { "x": 24, "y": 964 }, "mixtral_top2.expert7.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert7.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert7.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert7.expert.down": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "mixtral_top2" ], "selectedNodeIds": [ "mixtral_top2.router" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "mixtral_top2.router" }, "snapshotId": "snapshot-82f8c493b242ec59", "exportedAt": "2026-09-17T06:02:58.595933+00:00", "modelName": "SparseRoutingComparison", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.router.logits\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.router.noise_scale\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert3\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert4\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert5\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert6\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert7\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router.noise_scale\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"softplus@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert1#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert2#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert3#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert4#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert5#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert6#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert7#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"softplus@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[\"{\\\"id\\\":\\\"add@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.add\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.router.logits\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.router.noise_scale\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.mul\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.router#1\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.router#2\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"softplus@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Softplus\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\"],\"node\":{\"id\":\"mixtral_top2.router\",\"kind\":\"module\",\"op\":\"Module\",\"params\":{\"arg_0\":2,\"arg_1\":true,\"arg_2\":false},\"parent\":\"mixtral_top2\"}}" } ] }, { "id": "dispatch", "title": "Send only selected tokens to an expert", "note": "Dispatch gathers tokens with a nonzero gate for expert 0, evaluates its SwiGLU network, multiplies by the gate and scatters the result back. Unselected tokens never enter this expert's linear layers.", "view": { "camera": { "x": 867.5384615384617, "y": -240.47552447552448, "z": 0.5181159420289855 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "noisy_topk": { "x": 340, "y": 160 }, "switch_top1": { "x": 680, "y": 160 }, "mixtral_top2": { "x": 1020, "y": 160 }, "stack@SparseRoutingComparison#0": { "x": 1360, "y": 160 }, "output": { "x": 1700, "y": 160 }, "noisy_topk.router": { "x": 24, "y": 214 }, "noisy_topk.expert0": { "x": 24, "y": 364 }, "noisy_topk.expert1": { "x": 24, "y": 514 }, "add@noisy_topk#0": { "x": 24, "y": 664 }, "noisy_topk.expert2": { "x": 24, "y": 814 }, "add@noisy_topk#1": { "x": 24, "y": 964 }, "noisy_topk.expert3": { "x": 24, "y": 1114 }, "add@noisy_topk#2": { "x": 24, "y": 1264 }, "noisy_topk.expert4": { "x": 24, "y": 1414 }, "add@noisy_topk#3": { "x": 24, "y": 1564 }, "noisy_topk.expert5": { "x": 24, "y": 1714 }, "add@noisy_topk#4": { "x": 24, "y": 1864 }, "noisy_topk.expert6": { "x": 24, "y": 2014 }, "add@noisy_topk#5": { "x": 24, "y": 2164 }, "noisy_topk.expert7": { "x": 24, "y": 2314 }, "add@noisy_topk#6": { "x": 24, "y": 2464 }, "noisy_topk.router.logits": { "x": 24, "y": 214 }, "noisy_topk.router.noise_scale": { "x": 24, "y": 364 }, "softplus@noisy_topk.router#0": { "x": 24, "y": 514 }, "mul@noisy_topk.router#0": { "x": 24, "y": 664 }, "add@noisy_topk.router#0": { "x": 24, "y": 814 }, "python@noisy_topk.router#0": { "x": 24, "y": 964 }, "python@noisy_topk.router#1": { "x": 24, "y": 1114 }, "python@noisy_topk.router#2": { "x": 24, "y": 1264 }, "flatten@noisy_topk.expert0#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert0#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert0#1": { "x": 24, "y": 514 }, "noisy_topk.expert0.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert0#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert0#2": { "x": 24, "y": 964 }, "noisy_topk.expert0.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert0.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert0.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert1#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert1#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert1#1": { "x": 24, "y": 514 }, "noisy_topk.expert1.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert1#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert1#2": { "x": 24, "y": 964 }, "noisy_topk.expert1.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert1.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert1.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert2#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert2#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert2#1": { "x": 24, "y": 514 }, "noisy_topk.expert2.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert2#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert2#2": { "x": 24, "y": 964 }, "noisy_topk.expert2.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert2.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert2.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert3#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert3#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert3#1": { "x": 24, "y": 514 }, "noisy_topk.expert3.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert3#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert3#2": { "x": 24, "y": 964 }, "noisy_topk.expert3.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert3.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert3.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert4#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert4#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert4#1": { "x": 24, "y": 514 }, "noisy_topk.expert4.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert4#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert4#2": { "x": 24, "y": 964 }, "noisy_topk.expert4.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert4.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert4.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert5#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert5#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert5#1": { "x": 24, "y": 514 }, "noisy_topk.expert5.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert5#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert5#2": { "x": 24, "y": 964 }, "noisy_topk.expert5.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert5.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert5.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert6#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert6#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert6#1": { "x": 24, "y": 514 }, "noisy_topk.expert6.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert6#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert6#2": { "x": 24, "y": 964 }, "noisy_topk.expert6.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert6.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert6.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert7#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert7#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert7#1": { "x": 24, "y": 514 }, "noisy_topk.expert7.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert7#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert7#2": { "x": 24, "y": 964 }, "noisy_topk.expert7.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert7.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert7.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert7.expert.down": { "x": 24, "y": 814 }, "switch_top1.router": { "x": 24, "y": 214 }, "switch_top1.expert0": { "x": 24, "y": 364 }, "switch_top1.expert1": { "x": 24, "y": 514 }, "add@switch_top1#0": { "x": 24, "y": 664 }, "switch_top1.expert2": { "x": 24, "y": 814 }, "add@switch_top1#1": { "x": 24, "y": 964 }, "switch_top1.expert3": { "x": 24, "y": 1114 }, "add@switch_top1#2": { "x": 24, "y": 1264 }, "switch_top1.expert4": { "x": 24, "y": 1414 }, "add@switch_top1#3": { "x": 24, "y": 1564 }, "switch_top1.expert5": { "x": 24, "y": 1714 }, "add@switch_top1#4": { "x": 24, "y": 1864 }, "switch_top1.expert6": { "x": 24, "y": 2014 }, "add@switch_top1#5": { "x": 24, "y": 2164 }, "switch_top1.expert7": { "x": 24, "y": 2314 }, "add@switch_top1#6": { "x": 24, "y": 2464 }, "switch_top1.router.logits": { "x": 24, "y": 214 }, "switch_top1.router.noise_scale": { "x": 24, "y": 364 }, "softplus@switch_top1.router#0": { "x": 24, "y": 514 }, "mul@switch_top1.router#0": { "x": 24, "y": 664 }, "add@switch_top1.router#0": { "x": 24, "y": 814 }, "python@switch_top1.router#0": { "x": 24, "y": 964 }, "python@switch_top1.router#1": { "x": 24, "y": 1114 }, "python@switch_top1.router#2": { "x": 24, "y": 1264 }, "flatten@switch_top1.expert0#0": { "x": 24, "y": 214 }, "python@switch_top1.expert0#0": { "x": 24, "y": 364 }, "python@switch_top1.expert0#1": { "x": 24, "y": 514 }, "switch_top1.expert0.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert0#0": { "x": 24, "y": 814 }, "python@switch_top1.expert0#2": { "x": 24, "y": 964 }, "switch_top1.expert0.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert0.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert0.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert1#0": { "x": 24, "y": 214 }, "python@switch_top1.expert1#0": { "x": 24, "y": 364 }, "python@switch_top1.expert1#1": { "x": 24, "y": 514 }, "switch_top1.expert1.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert1#0": { "x": 24, "y": 814 }, "python@switch_top1.expert1#2": { "x": 24, "y": 964 }, "switch_top1.expert1.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert1.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert1.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert2#0": { "x": 24, "y": 214 }, "python@switch_top1.expert2#0": { "x": 24, "y": 364 }, "python@switch_top1.expert2#1": { "x": 24, "y": 514 }, "switch_top1.expert2.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert2#0": { "x": 24, "y": 814 }, "python@switch_top1.expert2#2": { "x": 24, "y": 964 }, "switch_top1.expert2.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert2.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert2.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert3#0": { "x": 24, "y": 214 }, "python@switch_top1.expert3#0": { "x": 24, "y": 364 }, "python@switch_top1.expert3#1": { "x": 24, "y": 514 }, "switch_top1.expert3.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert3#0": { "x": 24, "y": 814 }, "python@switch_top1.expert3#2": { "x": 24, "y": 964 }, "switch_top1.expert3.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert3.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert3.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert4#0": { "x": 24, "y": 214 }, "python@switch_top1.expert4#0": { "x": 24, "y": 364 }, "python@switch_top1.expert4#1": { "x": 24, "y": 514 }, "switch_top1.expert4.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert4#0": { "x": 24, "y": 814 }, "python@switch_top1.expert4#2": { "x": 24, "y": 964 }, "switch_top1.expert4.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert4.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert4.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert5#0": { "x": 24, "y": 214 }, "python@switch_top1.expert5#0": { "x": 24, "y": 364 }, "python@switch_top1.expert5#1": { "x": 24, "y": 514 }, "switch_top1.expert5.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert5#0": { "x": 24, "y": 814 }, "python@switch_top1.expert5#2": { "x": 24, "y": 964 }, "switch_top1.expert5.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert5.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert5.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert6#0": { "x": 24, "y": 214 }, "python@switch_top1.expert6#0": { "x": 24, "y": 364 }, "python@switch_top1.expert6#1": { "x": 24, "y": 514 }, "switch_top1.expert6.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert6#0": { "x": 24, "y": 814 }, "python@switch_top1.expert6#2": { "x": 24, "y": 964 }, "switch_top1.expert6.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert6.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert6.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert7#0": { "x": 24, "y": 214 }, "python@switch_top1.expert7#0": { "x": 24, "y": 364 }, "python@switch_top1.expert7#1": { "x": 24, "y": 514 }, "switch_top1.expert7.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert7#0": { "x": 24, "y": 814 }, "python@switch_top1.expert7#2": { "x": 24, "y": 964 }, "switch_top1.expert7.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert7.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert7.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert7.expert.down": { "x": 24, "y": 814 }, "mixtral_top2.router": { "x": 24, "y": 214 }, "mixtral_top2.expert0": { "x": 24, "y": 364 }, "mixtral_top2.expert1": { "x": 24, "y": 514 }, "add@mixtral_top2#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2": { "x": 24, "y": 814 }, "add@mixtral_top2#1": { "x": 24, "y": 964 }, "mixtral_top2.expert3": { "x": 24, "y": 1114 }, "add@mixtral_top2#2": { "x": 24, "y": 1264 }, "mixtral_top2.expert4": { "x": 24, "y": 1414 }, "add@mixtral_top2#3": { "x": 24, "y": 1564 }, "mixtral_top2.expert5": { "x": 24, "y": 1714 }, "add@mixtral_top2#4": { "x": 24, "y": 1864 }, "mixtral_top2.expert6": { "x": 24, "y": 2014 }, "add@mixtral_top2#5": { "x": 24, "y": 2164 }, "mixtral_top2.expert7": { "x": 24, "y": 2314 }, "add@mixtral_top2#6": { "x": 24, "y": 2464 }, "mixtral_top2.router.logits": { "x": 24, "y": 214 }, "mixtral_top2.router.noise_scale": { "x": 24, "y": 364 }, "softplus@mixtral_top2.router#0": { "x": 24, "y": 514 }, "mul@mixtral_top2.router#0": { "x": 24, "y": 664 }, "add@mixtral_top2.router#0": { "x": 24, "y": 814 }, "python@mixtral_top2.router#0": { "x": 24, "y": 964 }, "python@mixtral_top2.router#1": { "x": 24, "y": 1114 }, "python@mixtral_top2.router#2": { "x": 24, "y": 1264 }, "flatten@mixtral_top2.expert0#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert0#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert0#1": { "x": 24, "y": 514 }, "mixtral_top2.expert0.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert0#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert0#2": { "x": 24, "y": 964 }, "mixtral_top2.expert0.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert0.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert0.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert1#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert1#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert1#1": { "x": 24, "y": 514 }, "mixtral_top2.expert1.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert1#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert1#2": { "x": 24, "y": 964 }, "mixtral_top2.expert1.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert1.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert1.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert2#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert2#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert2#1": { "x": 24, "y": 514 }, "mixtral_top2.expert2.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert2#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert2#2": { "x": 24, "y": 964 }, "mixtral_top2.expert2.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert2.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert2.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert3#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert3#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert3#1": { "x": 24, "y": 514 }, "mixtral_top2.expert3.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert3#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert3#2": { "x": 24, "y": 964 }, "mixtral_top2.expert3.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert3.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert3.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert4#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert4#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert4#1": { "x": 24, "y": 514 }, "mixtral_top2.expert4.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert4#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert4#2": { "x": 24, "y": 964 }, "mixtral_top2.expert4.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert4.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert4.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert5#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert5#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert5#1": { "x": 24, "y": 514 }, "mixtral_top2.expert5.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert5#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert5#2": { "x": 24, "y": 964 }, "mixtral_top2.expert5.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert5.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert5.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert6#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert6#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert6#1": { "x": 24, "y": 514 }, "mixtral_top2.expert6.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert6#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert6#2": { "x": 24, "y": 964 }, "mixtral_top2.expert6.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert6.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert6.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert7#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert7#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert7#1": { "x": 24, "y": 514 }, "mixtral_top2.expert7.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert7#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert7#2": { "x": 24, "y": 964 }, "mixtral_top2.expert7.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert7.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert7.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert7.expert.down": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "mixtral_top2", "mixtral_top2.expert0" ], "selectedNodeIds": [ "mixtral_top2.expert0" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "mixtral_top2.expert0" }, "snapshotId": "snapshot-82f8c493b242ec59", "exportedAt": "2026-09-17T06:02:58.595933+00:00", "modelName": "SparseRoutingComparison", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.expert0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@mixtral_top2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.expert0.expert\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.expert0.expert.down\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.expert0.expert.gate\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0.expert.silu\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.expert0.expert.silu\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0.expert#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.expert0.expert.up\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0.expert#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mul@mixtral_top2.expert0.expert#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0.expert.down\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.expert0#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.expert0#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.expert0#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0.expert\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.expert0#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0.expert.gate\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.expert0#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0.expert.up\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@mixtral_top2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[\"{\\\"id\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Flatten\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.expert0.expert\\\",\\\"kind\\\":\\\"module\\\",\\\"op\\\":\\\"Module\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.expert0.expert.down\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":12,\\\"out_features\\\":8},\\\"parent\\\":\\\"mixtral_top2.expert0.expert\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.expert0.expert.gate\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":12},\\\"parent\\\":\\\"mixtral_top2.expert0.expert\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.expert0.expert.silu\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"SiLU\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0.expert\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.expert0.expert.up\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":12},\\\"parent\\\":\\\"mixtral_top2.expert0.expert\\\"}\",\"{\\\"id\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.mul\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0\\\"}\",\"{\\\"id\\\":\\\"mul@mixtral_top2.expert0.expert#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.mul\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0.expert\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.expert0#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.expert0#1\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.expert0\\\"}\"],\"node\":{\"id\":\"mixtral_top2.expert0\",\"kind\":\"module\",\"op\":\"Module\",\"params\":{\"arg_0\":0},\"parent\":\"mixtral_top2\"}}" } ] }, { "id": "switch", "title": "Compare Switch-style top-1", "note": "Switch sends each token to one expert and retains the winning full-softmax probability as its output weight. This differs from normalizing a top-1 gate to one. The simplified block omits capacity limits and auxiliary-loss training.", "view": { "camera": { "x": 291, "y": -4, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "noisy_topk": { "x": 340, "y": 160 }, "switch_top1": { "x": 680, "y": 160 }, "mixtral_top2": { "x": 1020, "y": 160 }, "stack@SparseRoutingComparison#0": { "x": 1360, "y": 160 }, "output": { "x": 1700, "y": 160 }, "noisy_topk.router": { "x": 24, "y": 214 }, "noisy_topk.expert0": { "x": 24, "y": 364 }, "noisy_topk.expert1": { "x": 24, "y": 514 }, "add@noisy_topk#0": { "x": 24, "y": 664 }, "noisy_topk.expert2": { "x": 24, "y": 814 }, "add@noisy_topk#1": { "x": 24, "y": 964 }, "noisy_topk.expert3": { "x": 24, "y": 1114 }, "add@noisy_topk#2": { "x": 24, "y": 1264 }, "noisy_topk.expert4": { "x": 24, "y": 1414 }, "add@noisy_topk#3": { "x": 24, "y": 1564 }, "noisy_topk.expert5": { "x": 24, "y": 1714 }, "add@noisy_topk#4": { "x": 24, "y": 1864 }, "noisy_topk.expert6": { "x": 24, "y": 2014 }, "add@noisy_topk#5": { "x": 24, "y": 2164 }, "noisy_topk.expert7": { "x": 24, "y": 2314 }, "add@noisy_topk#6": { "x": 24, "y": 2464 }, "noisy_topk.router.logits": { "x": 24, "y": 214 }, "noisy_topk.router.noise_scale": { "x": 24, "y": 364 }, "softplus@noisy_topk.router#0": { "x": 24, "y": 514 }, "mul@noisy_topk.router#0": { "x": 24, "y": 664 }, "add@noisy_topk.router#0": { "x": 24, "y": 814 }, "python@noisy_topk.router#0": { "x": 24, "y": 964 }, "python@noisy_topk.router#1": { "x": 24, "y": 1114 }, "python@noisy_topk.router#2": { "x": 24, "y": 1264 }, "flatten@noisy_topk.expert0#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert0#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert0#1": { "x": 24, "y": 514 }, "noisy_topk.expert0.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert0#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert0#2": { "x": 24, "y": 964 }, "noisy_topk.expert0.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert0.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert0.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert1#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert1#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert1#1": { "x": 24, "y": 514 }, "noisy_topk.expert1.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert1#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert1#2": { "x": 24, "y": 964 }, "noisy_topk.expert1.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert1.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert1.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert2#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert2#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert2#1": { "x": 24, "y": 514 }, "noisy_topk.expert2.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert2#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert2#2": { "x": 24, "y": 964 }, "noisy_topk.expert2.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert2.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert2.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert3#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert3#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert3#1": { "x": 24, "y": 514 }, "noisy_topk.expert3.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert3#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert3#2": { "x": 24, "y": 964 }, "noisy_topk.expert3.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert3.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert3.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert4#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert4#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert4#1": { "x": 24, "y": 514 }, "noisy_topk.expert4.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert4#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert4#2": { "x": 24, "y": 964 }, "noisy_topk.expert4.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert4.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert4.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert5#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert5#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert5#1": { "x": 24, "y": 514 }, "noisy_topk.expert5.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert5#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert5#2": { "x": 24, "y": 964 }, "noisy_topk.expert5.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert5.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert5.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert6#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert6#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert6#1": { "x": 24, "y": 514 }, "noisy_topk.expert6.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert6#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert6#2": { "x": 24, "y": 964 }, "noisy_topk.expert6.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert6.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert6.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert7#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert7#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert7#1": { "x": 24, "y": 514 }, "noisy_topk.expert7.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert7#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert7#2": { "x": 24, "y": 964 }, "noisy_topk.expert7.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert7.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert7.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert7.expert.down": { "x": 24, "y": 814 }, "switch_top1.router": { "x": 24, "y": 214 }, "switch_top1.expert0": { "x": 24, "y": 364 }, "switch_top1.expert1": { "x": 24, "y": 514 }, "add@switch_top1#0": { "x": 24, "y": 664 }, "switch_top1.expert2": { "x": 24, "y": 814 }, "add@switch_top1#1": { "x": 24, "y": 964 }, "switch_top1.expert3": { "x": 24, "y": 1114 }, "add@switch_top1#2": { "x": 24, "y": 1264 }, "switch_top1.expert4": { "x": 24, "y": 1414 }, "add@switch_top1#3": { "x": 24, "y": 1564 }, "switch_top1.expert5": { "x": 24, "y": 1714 }, "add@switch_top1#4": { "x": 24, "y": 1864 }, "switch_top1.expert6": { "x": 24, "y": 2014 }, "add@switch_top1#5": { "x": 24, "y": 2164 }, "switch_top1.expert7": { "x": 24, "y": 2314 }, "add@switch_top1#6": { "x": 24, "y": 2464 }, "switch_top1.router.logits": { "x": 24, "y": 214 }, "switch_top1.router.noise_scale": { "x": 24, "y": 364 }, "softplus@switch_top1.router#0": { "x": 24, "y": 514 }, "mul@switch_top1.router#0": { "x": 24, "y": 664 }, "add@switch_top1.router#0": { "x": 24, "y": 814 }, "python@switch_top1.router#0": { "x": 24, "y": 964 }, "python@switch_top1.router#1": { "x": 24, "y": 1114 }, "python@switch_top1.router#2": { "x": 24, "y": 1264 }, "flatten@switch_top1.expert0#0": { "x": 24, "y": 214 }, "python@switch_top1.expert0#0": { "x": 24, "y": 364 }, "python@switch_top1.expert0#1": { "x": 24, "y": 514 }, "switch_top1.expert0.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert0#0": { "x": 24, "y": 814 }, "python@switch_top1.expert0#2": { "x": 24, "y": 964 }, "switch_top1.expert0.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert0.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert0.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert1#0": { "x": 24, "y": 214 }, "python@switch_top1.expert1#0": { "x": 24, "y": 364 }, "python@switch_top1.expert1#1": { "x": 24, "y": 514 }, "switch_top1.expert1.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert1#0": { "x": 24, "y": 814 }, "python@switch_top1.expert1#2": { "x": 24, "y": 964 }, "switch_top1.expert1.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert1.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert1.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert2#0": { "x": 24, "y": 214 }, "python@switch_top1.expert2#0": { "x": 24, "y": 364 }, "python@switch_top1.expert2#1": { "x": 24, "y": 514 }, "switch_top1.expert2.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert2#0": { "x": 24, "y": 814 }, "python@switch_top1.expert2#2": { "x": 24, "y": 964 }, "switch_top1.expert2.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert2.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert2.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert3#0": { "x": 24, "y": 214 }, "python@switch_top1.expert3#0": { "x": 24, "y": 364 }, "python@switch_top1.expert3#1": { "x": 24, "y": 514 }, "switch_top1.expert3.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert3#0": { "x": 24, "y": 814 }, "python@switch_top1.expert3#2": { "x": 24, "y": 964 }, "switch_top1.expert3.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert3.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert3.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert4#0": { "x": 24, "y": 214 }, "python@switch_top1.expert4#0": { "x": 24, "y": 364 }, "python@switch_top1.expert4#1": { "x": 24, "y": 514 }, "switch_top1.expert4.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert4#0": { "x": 24, "y": 814 }, "python@switch_top1.expert4#2": { "x": 24, "y": 964 }, "switch_top1.expert4.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert4.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert4.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert5#0": { "x": 24, "y": 214 }, "python@switch_top1.expert5#0": { "x": 24, "y": 364 }, "python@switch_top1.expert5#1": { "x": 24, "y": 514 }, "switch_top1.expert5.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert5#0": { "x": 24, "y": 814 }, "python@switch_top1.expert5#2": { "x": 24, "y": 964 }, "switch_top1.expert5.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert5.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert5.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert6#0": { "x": 24, "y": 214 }, "python@switch_top1.expert6#0": { "x": 24, "y": 364 }, "python@switch_top1.expert6#1": { "x": 24, "y": 514 }, "switch_top1.expert6.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert6#0": { "x": 24, "y": 814 }, "python@switch_top1.expert6#2": { "x": 24, "y": 964 }, "switch_top1.expert6.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert6.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert6.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert7#0": { "x": 24, "y": 214 }, "python@switch_top1.expert7#0": { "x": 24, "y": 364 }, "python@switch_top1.expert7#1": { "x": 24, "y": 514 }, "switch_top1.expert7.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert7#0": { "x": 24, "y": 814 }, "python@switch_top1.expert7#2": { "x": 24, "y": 964 }, "switch_top1.expert7.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert7.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert7.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert7.expert.down": { "x": 24, "y": 814 }, "mixtral_top2.router": { "x": 24, "y": 214 }, "mixtral_top2.expert0": { "x": 24, "y": 364 }, "mixtral_top2.expert1": { "x": 24, "y": 514 }, "add@mixtral_top2#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2": { "x": 24, "y": 814 }, "add@mixtral_top2#1": { "x": 24, "y": 964 }, "mixtral_top2.expert3": { "x": 24, "y": 1114 }, "add@mixtral_top2#2": { "x": 24, "y": 1264 }, "mixtral_top2.expert4": { "x": 24, "y": 1414 }, "add@mixtral_top2#3": { "x": 24, "y": 1564 }, "mixtral_top2.expert5": { "x": 24, "y": 1714 }, "add@mixtral_top2#4": { "x": 24, "y": 1864 }, "mixtral_top2.expert6": { "x": 24, "y": 2014 }, "add@mixtral_top2#5": { "x": 24, "y": 2164 }, "mixtral_top2.expert7": { "x": 24, "y": 2314 }, "add@mixtral_top2#6": { "x": 24, "y": 2464 }, "mixtral_top2.router.logits": { "x": 24, "y": 214 }, "mixtral_top2.router.noise_scale": { "x": 24, "y": 364 }, "softplus@mixtral_top2.router#0": { "x": 24, "y": 514 }, "mul@mixtral_top2.router#0": { "x": 24, "y": 664 }, "add@mixtral_top2.router#0": { "x": 24, "y": 814 }, "python@mixtral_top2.router#0": { "x": 24, "y": 964 }, "python@mixtral_top2.router#1": { "x": 24, "y": 1114 }, "python@mixtral_top2.router#2": { "x": 24, "y": 1264 }, "flatten@mixtral_top2.expert0#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert0#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert0#1": { "x": 24, "y": 514 }, "mixtral_top2.expert0.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert0#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert0#2": { "x": 24, "y": 964 }, "mixtral_top2.expert0.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert0.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert0.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert1#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert1#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert1#1": { "x": 24, "y": 514 }, "mixtral_top2.expert1.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert1#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert1#2": { "x": 24, "y": 964 }, "mixtral_top2.expert1.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert1.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert1.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert2#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert2#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert2#1": { "x": 24, "y": 514 }, "mixtral_top2.expert2.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert2#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert2#2": { "x": 24, "y": 964 }, "mixtral_top2.expert2.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert2.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert2.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert3#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert3#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert3#1": { "x": 24, "y": 514 }, "mixtral_top2.expert3.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert3#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert3#2": { "x": 24, "y": 964 }, "mixtral_top2.expert3.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert3.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert3.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert4#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert4#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert4#1": { "x": 24, "y": 514 }, "mixtral_top2.expert4.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert4#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert4#2": { "x": 24, "y": 964 }, "mixtral_top2.expert4.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert4.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert4.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert5#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert5#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert5#1": { "x": 24, "y": 514 }, "mixtral_top2.expert5.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert5#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert5#2": { "x": 24, "y": 964 }, "mixtral_top2.expert5.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert5.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert5.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert6#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert6#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert6#1": { "x": 24, "y": 514 }, "mixtral_top2.expert6.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert6#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert6#2": { "x": 24, "y": 964 }, "mixtral_top2.expert6.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert6.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert6.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert7#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert7#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert7#1": { "x": 24, "y": 514 }, "mixtral_top2.expert7.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert7#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert7#2": { "x": 24, "y": 964 }, "mixtral_top2.expert7.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert7.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert7.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert7.expert.down": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "switch_top1" ], "selectedNodeIds": [ "switch_top1.router" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "switch_top1.router" }, "snapshotId": "snapshot-82f8c493b242ec59", "exportedAt": "2026-09-17T06:02:58.595933+00:00", "modelName": "SparseRoutingComparison", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@switch_top1.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.router.logits\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.router.noise_scale\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mul@switch_top1.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@switch_top1.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.router#1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@switch_top1.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert1#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert2#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert3#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert4#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert5#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert6#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@switch_top1.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@switch_top1.expert7#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"softplus@switch_top1.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert3\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert4\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert5\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert6\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"switch_top1.expert7\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@switch_top1.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@switch_top1.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"switch_top1.router.noise_scale\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"softplus@switch_top1.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[\"{\\\"id\\\":\\\"add@switch_top1.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.add\\\",\\\"params\\\":{},\\\"parent\\\":\\\"switch_top1.router\\\"}\",\"{\\\"id\\\":\\\"mul@switch_top1.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.mul\\\",\\\"params\\\":{},\\\"parent\\\":\\\"switch_top1.router\\\"}\",\"{\\\"id\\\":\\\"python@switch_top1.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"switch_top1.router\\\"}\",\"{\\\"id\\\":\\\"python@switch_top1.router#1\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"switch_top1.router\\\"}\",\"{\\\"id\\\":\\\"python@switch_top1.router#2\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"switch_top1.router\\\"}\",\"{\\\"id\\\":\\\"softplus@switch_top1.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Softplus\\\",\\\"params\\\":{},\\\"parent\\\":\\\"switch_top1.router\\\"}\",\"{\\\"id\\\":\\\"switch_top1.router.logits\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"switch_top1.router\\\"}\",\"{\\\"id\\\":\\\"switch_top1.router.noise_scale\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"switch_top1.router\\\"}\"],\"node\":{\"id\":\"switch_top1.router\",\"kind\":\"module\",\"op\":\"Module\",\"params\":{\"arg_0\":2,\"arg_1\":true,\"arg_2\":false},\"parent\":\"switch_top1\"}}" } ] }, { "id": "noise", "title": "Explore the earlier noisy top-k route", "note": "The 2017 mechanism adds Gaussian noise scaled by softplus of a learned projection during training, then selects top-k. Evaluation disables noise. A seeded training case below lets you inspect the resulting changed routes.", "view": { "camera": { "x": 291, "y": -4, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "noisy_topk": { "x": 340, "y": 160 }, "switch_top1": { "x": 680, "y": 160 }, "mixtral_top2": { "x": 1020, "y": 160 }, "stack@SparseRoutingComparison#0": { "x": 1360, "y": 160 }, "output": { "x": 1700, "y": 160 }, "noisy_topk.router": { "x": 24, "y": 214 }, "noisy_topk.expert0": { "x": 24, "y": 364 }, "noisy_topk.expert1": { "x": 24, "y": 514 }, "add@noisy_topk#0": { "x": 24, "y": 664 }, "noisy_topk.expert2": { "x": 24, "y": 814 }, "add@noisy_topk#1": { "x": 24, "y": 964 }, "noisy_topk.expert3": { "x": 24, "y": 1114 }, "add@noisy_topk#2": { "x": 24, "y": 1264 }, "noisy_topk.expert4": { "x": 24, "y": 1414 }, "add@noisy_topk#3": { "x": 24, "y": 1564 }, "noisy_topk.expert5": { "x": 24, "y": 1714 }, "add@noisy_topk#4": { "x": 24, "y": 1864 }, "noisy_topk.expert6": { "x": 24, "y": 2014 }, "add@noisy_topk#5": { "x": 24, "y": 2164 }, "noisy_topk.expert7": { "x": 24, "y": 2314 }, "add@noisy_topk#6": { "x": 24, "y": 2464 }, "noisy_topk.router.logits": { "x": 24, "y": 214 }, "noisy_topk.router.noise_scale": { "x": 24, "y": 364 }, "softplus@noisy_topk.router#0": { "x": 24, "y": 514 }, "mul@noisy_topk.router#0": { "x": 24, "y": 664 }, "add@noisy_topk.router#0": { "x": 24, "y": 814 }, "python@noisy_topk.router#0": { "x": 24, "y": 964 }, "python@noisy_topk.router#1": { "x": 24, "y": 1114 }, "python@noisy_topk.router#2": { "x": 24, "y": 1264 }, "flatten@noisy_topk.expert0#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert0#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert0#1": { "x": 24, "y": 514 }, "noisy_topk.expert0.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert0#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert0#2": { "x": 24, "y": 964 }, "noisy_topk.expert0.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert0.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert0.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert1#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert1#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert1#1": { "x": 24, "y": 514 }, "noisy_topk.expert1.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert1#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert1#2": { "x": 24, "y": 964 }, "noisy_topk.expert1.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert1.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert1.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert2#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert2#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert2#1": { "x": 24, "y": 514 }, "noisy_topk.expert2.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert2#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert2#2": { "x": 24, "y": 964 }, "noisy_topk.expert2.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert2.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert2.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert3#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert3#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert3#1": { "x": 24, "y": 514 }, "noisy_topk.expert3.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert3#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert3#2": { "x": 24, "y": 964 }, "noisy_topk.expert3.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert3.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert3.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert4#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert4#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert4#1": { "x": 24, "y": 514 }, "noisy_topk.expert4.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert4#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert4#2": { "x": 24, "y": 964 }, "noisy_topk.expert4.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert4.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert4.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert5#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert5#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert5#1": { "x": 24, "y": 514 }, "noisy_topk.expert5.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert5#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert5#2": { "x": 24, "y": 964 }, "noisy_topk.expert5.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert5.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert5.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert6#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert6#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert6#1": { "x": 24, "y": 514 }, "noisy_topk.expert6.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert6#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert6#2": { "x": 24, "y": 964 }, "noisy_topk.expert6.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert6.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert6.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert7#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert7#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert7#1": { "x": 24, "y": 514 }, "noisy_topk.expert7.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert7#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert7#2": { "x": 24, "y": 964 }, "noisy_topk.expert7.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert7.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert7.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert7.expert.down": { "x": 24, "y": 814 }, "switch_top1.router": { "x": 24, "y": 214 }, "switch_top1.expert0": { "x": 24, "y": 364 }, "switch_top1.expert1": { "x": 24, "y": 514 }, "add@switch_top1#0": { "x": 24, "y": 664 }, "switch_top1.expert2": { "x": 24, "y": 814 }, "add@switch_top1#1": { "x": 24, "y": 964 }, "switch_top1.expert3": { "x": 24, "y": 1114 }, "add@switch_top1#2": { "x": 24, "y": 1264 }, "switch_top1.expert4": { "x": 24, "y": 1414 }, "add@switch_top1#3": { "x": 24, "y": 1564 }, "switch_top1.expert5": { "x": 24, "y": 1714 }, "add@switch_top1#4": { "x": 24, "y": 1864 }, "switch_top1.expert6": { "x": 24, "y": 2014 }, "add@switch_top1#5": { "x": 24, "y": 2164 }, "switch_top1.expert7": { "x": 24, "y": 2314 }, "add@switch_top1#6": { "x": 24, "y": 2464 }, "switch_top1.router.logits": { "x": 24, "y": 214 }, "switch_top1.router.noise_scale": { "x": 24, "y": 364 }, "softplus@switch_top1.router#0": { "x": 24, "y": 514 }, "mul@switch_top1.router#0": { "x": 24, "y": 664 }, "add@switch_top1.router#0": { "x": 24, "y": 814 }, "python@switch_top1.router#0": { "x": 24, "y": 964 }, "python@switch_top1.router#1": { "x": 24, "y": 1114 }, "python@switch_top1.router#2": { "x": 24, "y": 1264 }, "flatten@switch_top1.expert0#0": { "x": 24, "y": 214 }, "python@switch_top1.expert0#0": { "x": 24, "y": 364 }, "python@switch_top1.expert0#1": { "x": 24, "y": 514 }, "switch_top1.expert0.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert0#0": { "x": 24, "y": 814 }, "python@switch_top1.expert0#2": { "x": 24, "y": 964 }, "switch_top1.expert0.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert0.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert0.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert1#0": { "x": 24, "y": 214 }, "python@switch_top1.expert1#0": { "x": 24, "y": 364 }, "python@switch_top1.expert1#1": { "x": 24, "y": 514 }, "switch_top1.expert1.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert1#0": { "x": 24, "y": 814 }, "python@switch_top1.expert1#2": { "x": 24, "y": 964 }, "switch_top1.expert1.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert1.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert1.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert2#0": { "x": 24, "y": 214 }, "python@switch_top1.expert2#0": { "x": 24, "y": 364 }, "python@switch_top1.expert2#1": { "x": 24, "y": 514 }, "switch_top1.expert2.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert2#0": { "x": 24, "y": 814 }, "python@switch_top1.expert2#2": { "x": 24, "y": 964 }, "switch_top1.expert2.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert2.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert2.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert3#0": { "x": 24, "y": 214 }, "python@switch_top1.expert3#0": { "x": 24, "y": 364 }, "python@switch_top1.expert3#1": { "x": 24, "y": 514 }, "switch_top1.expert3.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert3#0": { "x": 24, "y": 814 }, "python@switch_top1.expert3#2": { "x": 24, "y": 964 }, "switch_top1.expert3.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert3.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert3.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert4#0": { "x": 24, "y": 214 }, "python@switch_top1.expert4#0": { "x": 24, "y": 364 }, "python@switch_top1.expert4#1": { "x": 24, "y": 514 }, "switch_top1.expert4.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert4#0": { "x": 24, "y": 814 }, "python@switch_top1.expert4#2": { "x": 24, "y": 964 }, "switch_top1.expert4.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert4.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert4.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert5#0": { "x": 24, "y": 214 }, "python@switch_top1.expert5#0": { "x": 24, "y": 364 }, "python@switch_top1.expert5#1": { "x": 24, "y": 514 }, "switch_top1.expert5.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert5#0": { "x": 24, "y": 814 }, "python@switch_top1.expert5#2": { "x": 24, "y": 964 }, "switch_top1.expert5.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert5.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert5.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert6#0": { "x": 24, "y": 214 }, "python@switch_top1.expert6#0": { "x": 24, "y": 364 }, "python@switch_top1.expert6#1": { "x": 24, "y": 514 }, "switch_top1.expert6.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert6#0": { "x": 24, "y": 814 }, "python@switch_top1.expert6#2": { "x": 24, "y": 964 }, "switch_top1.expert6.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert6.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert6.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert7#0": { "x": 24, "y": 214 }, "python@switch_top1.expert7#0": { "x": 24, "y": 364 }, "python@switch_top1.expert7#1": { "x": 24, "y": 514 }, "switch_top1.expert7.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert7#0": { "x": 24, "y": 814 }, "python@switch_top1.expert7#2": { "x": 24, "y": 964 }, "switch_top1.expert7.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert7.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert7.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert7.expert.down": { "x": 24, "y": 814 }, "mixtral_top2.router": { "x": 24, "y": 214 }, "mixtral_top2.expert0": { "x": 24, "y": 364 }, "mixtral_top2.expert1": { "x": 24, "y": 514 }, "add@mixtral_top2#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2": { "x": 24, "y": 814 }, "add@mixtral_top2#1": { "x": 24, "y": 964 }, "mixtral_top2.expert3": { "x": 24, "y": 1114 }, "add@mixtral_top2#2": { "x": 24, "y": 1264 }, "mixtral_top2.expert4": { "x": 24, "y": 1414 }, "add@mixtral_top2#3": { "x": 24, "y": 1564 }, "mixtral_top2.expert5": { "x": 24, "y": 1714 }, "add@mixtral_top2#4": { "x": 24, "y": 1864 }, "mixtral_top2.expert6": { "x": 24, "y": 2014 }, "add@mixtral_top2#5": { "x": 24, "y": 2164 }, "mixtral_top2.expert7": { "x": 24, "y": 2314 }, "add@mixtral_top2#6": { "x": 24, "y": 2464 }, "mixtral_top2.router.logits": { "x": 24, "y": 214 }, "mixtral_top2.router.noise_scale": { "x": 24, "y": 364 }, "softplus@mixtral_top2.router#0": { "x": 24, "y": 514 }, "mul@mixtral_top2.router#0": { "x": 24, "y": 664 }, "add@mixtral_top2.router#0": { "x": 24, "y": 814 }, "python@mixtral_top2.router#0": { "x": 24, "y": 964 }, "python@mixtral_top2.router#1": { "x": 24, "y": 1114 }, "python@mixtral_top2.router#2": { "x": 24, "y": 1264 }, "flatten@mixtral_top2.expert0#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert0#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert0#1": { "x": 24, "y": 514 }, "mixtral_top2.expert0.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert0#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert0#2": { "x": 24, "y": 964 }, "mixtral_top2.expert0.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert0.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert0.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert1#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert1#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert1#1": { "x": 24, "y": 514 }, "mixtral_top2.expert1.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert1#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert1#2": { "x": 24, "y": 964 }, "mixtral_top2.expert1.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert1.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert1.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert2#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert2#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert2#1": { "x": 24, "y": 514 }, "mixtral_top2.expert2.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert2#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert2#2": { "x": 24, "y": 964 }, "mixtral_top2.expert2.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert2.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert2.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert3#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert3#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert3#1": { "x": 24, "y": 514 }, "mixtral_top2.expert3.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert3#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert3#2": { "x": 24, "y": 964 }, "mixtral_top2.expert3.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert3.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert3.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert4#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert4#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert4#1": { "x": 24, "y": 514 }, "mixtral_top2.expert4.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert4#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert4#2": { "x": 24, "y": 964 }, "mixtral_top2.expert4.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert4.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert4.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert5#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert5#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert5#1": { "x": 24, "y": 514 }, "mixtral_top2.expert5.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert5#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert5#2": { "x": 24, "y": 964 }, "mixtral_top2.expert5.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert5.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert5.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert6#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert6#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert6#1": { "x": 24, "y": 514 }, "mixtral_top2.expert6.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert6#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert6#2": { "x": 24, "y": 964 }, "mixtral_top2.expert6.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert6.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert6.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert7#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert7#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert7#1": { "x": 24, "y": 514 }, "mixtral_top2.expert7.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert7#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert7#2": { "x": 24, "y": 964 }, "mixtral_top2.expert7.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert7.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert7.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert7.expert.down": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "noisy_topk" ], "selectedNodeIds": [ "noisy_topk.router" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "noisy_topk.router" }, "snapshotId": "snapshot-82f8c493b242ec59", "exportedAt": "2026-09-17T06:02:58.595933+00:00", "modelName": "SparseRoutingComparison", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@noisy_topk.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.router.logits\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.router.noise_scale\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mul@noisy_topk.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@noisy_topk.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert3\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert4\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert5\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert6\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"noisy_topk.expert7\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@noisy_topk.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"noisy_topk.router.noise_scale\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"softplus@noisy_topk.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.router#1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@noisy_topk.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert1#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert2#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert3#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert4#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert5#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert6#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@noisy_topk.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@noisy_topk.expert7#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"softplus@noisy_topk.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@noisy_topk.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[\"{\\\"id\\\":\\\"add@noisy_topk.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.add\\\",\\\"params\\\":{},\\\"parent\\\":\\\"noisy_topk.router\\\"}\",\"{\\\"id\\\":\\\"mul@noisy_topk.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.mul\\\",\\\"params\\\":{},\\\"parent\\\":\\\"noisy_topk.router\\\"}\",\"{\\\"id\\\":\\\"noisy_topk.router.logits\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"noisy_topk.router\\\"}\",\"{\\\"id\\\":\\\"noisy_topk.router.noise_scale\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"noisy_topk.router\\\"}\",\"{\\\"id\\\":\\\"python@noisy_topk.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"noisy_topk.router\\\"}\",\"{\\\"id\\\":\\\"python@noisy_topk.router#1\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"noisy_topk.router\\\"}\",\"{\\\"id\\\":\\\"python@noisy_topk.router#2\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"noisy_topk.router\\\"}\",\"{\\\"id\\\":\\\"softplus@noisy_topk.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Softplus\\\",\\\"params\\\":{},\\\"parent\\\":\\\"noisy_topk.router\\\"}\"],\"node\":{\"id\":\"noisy_topk.router\",\"kind\":\"module\",\"op\":\"Module\",\"params\":{\"arg_0\":2,\"arg_1\":true,\"arg_2\":false},\"parent\":\"noisy_topk\"}}" } ] } ] }, "threads": [], "savedViews": [ { "id": "route", "title": "Score and select two experts", "note": "The router scores eight experts, selects the two largest probabilities and renormalizes those selected weights. Each token can choose a different pair. The recorded diagram shows actual routes for synthetic inputs.", "view": { "camera": { "x": 291, "y": -4, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "noisy_topk": { "x": 340, "y": 160 }, "switch_top1": { "x": 680, "y": 160 }, "mixtral_top2": { "x": 1020, "y": 160 }, "stack@SparseRoutingComparison#0": { "x": 1360, "y": 160 }, "output": { "x": 1700, "y": 160 }, "noisy_topk.router": { "x": 24, "y": 214 }, "noisy_topk.expert0": { "x": 24, "y": 364 }, "noisy_topk.expert1": { "x": 24, "y": 514 }, "add@noisy_topk#0": { "x": 24, "y": 664 }, "noisy_topk.expert2": { "x": 24, "y": 814 }, "add@noisy_topk#1": { "x": 24, "y": 964 }, "noisy_topk.expert3": { "x": 24, "y": 1114 }, "add@noisy_topk#2": { "x": 24, "y": 1264 }, "noisy_topk.expert4": { "x": 24, "y": 1414 }, "add@noisy_topk#3": { "x": 24, "y": 1564 }, "noisy_topk.expert5": { "x": 24, "y": 1714 }, "add@noisy_topk#4": { "x": 24, "y": 1864 }, "noisy_topk.expert6": { "x": 24, "y": 2014 }, "add@noisy_topk#5": { "x": 24, "y": 2164 }, "noisy_topk.expert7": { "x": 24, "y": 2314 }, "add@noisy_topk#6": { "x": 24, "y": 2464 }, "noisy_topk.router.logits": { "x": 24, "y": 214 }, "noisy_topk.router.noise_scale": { "x": 24, "y": 364 }, "softplus@noisy_topk.router#0": { "x": 24, "y": 514 }, "mul@noisy_topk.router#0": { "x": 24, "y": 664 }, "add@noisy_topk.router#0": { "x": 24, "y": 814 }, "python@noisy_topk.router#0": { "x": 24, "y": 964 }, "python@noisy_topk.router#1": { "x": 24, "y": 1114 }, "python@noisy_topk.router#2": { "x": 24, "y": 1264 }, "flatten@noisy_topk.expert0#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert0#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert0#1": { "x": 24, "y": 514 }, "noisy_topk.expert0.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert0#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert0#2": { "x": 24, "y": 964 }, "noisy_topk.expert0.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert0.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert0.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert1#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert1#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert1#1": { "x": 24, "y": 514 }, "noisy_topk.expert1.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert1#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert1#2": { "x": 24, "y": 964 }, "noisy_topk.expert1.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert1.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert1.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert2#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert2#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert2#1": { "x": 24, "y": 514 }, "noisy_topk.expert2.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert2#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert2#2": { "x": 24, "y": 964 }, "noisy_topk.expert2.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert2.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert2.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert3#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert3#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert3#1": { "x": 24, "y": 514 }, "noisy_topk.expert3.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert3#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert3#2": { "x": 24, "y": 964 }, "noisy_topk.expert3.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert3.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert3.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert4#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert4#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert4#1": { "x": 24, "y": 514 }, "noisy_topk.expert4.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert4#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert4#2": { "x": 24, "y": 964 }, "noisy_topk.expert4.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert4.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert4.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert5#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert5#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert5#1": { "x": 24, "y": 514 }, "noisy_topk.expert5.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert5#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert5#2": { "x": 24, "y": 964 }, "noisy_topk.expert5.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert5.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert5.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert6#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert6#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert6#1": { "x": 24, "y": 514 }, "noisy_topk.expert6.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert6#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert6#2": { "x": 24, "y": 964 }, "noisy_topk.expert6.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert6.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert6.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@noisy_topk.expert7#0": { "x": 24, "y": 214 }, "python@noisy_topk.expert7#0": { "x": 24, "y": 364 }, "python@noisy_topk.expert7#1": { "x": 24, "y": 514 }, "noisy_topk.expert7.expert": { "x": 24, "y": 664 }, "mul@noisy_topk.expert7#0": { "x": 24, "y": 814 }, "python@noisy_topk.expert7#2": { "x": 24, "y": 964 }, "noisy_topk.expert7.expert.gate": { "x": 24, "y": 214 }, "noisy_topk.expert7.expert.up": { "x": 24, "y": 364 }, "noisy_topk.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@noisy_topk.expert7.expert#0": { "x": 24, "y": 664 }, "noisy_topk.expert7.expert.down": { "x": 24, "y": 814 }, "switch_top1.router": { "x": 24, "y": 214 }, "switch_top1.expert0": { "x": 24, "y": 364 }, "switch_top1.expert1": { "x": 24, "y": 514 }, "add@switch_top1#0": { "x": 24, "y": 664 }, "switch_top1.expert2": { "x": 24, "y": 814 }, "add@switch_top1#1": { "x": 24, "y": 964 }, "switch_top1.expert3": { "x": 24, "y": 1114 }, "add@switch_top1#2": { "x": 24, "y": 1264 }, "switch_top1.expert4": { "x": 24, "y": 1414 }, "add@switch_top1#3": { "x": 24, "y": 1564 }, "switch_top1.expert5": { "x": 24, "y": 1714 }, "add@switch_top1#4": { "x": 24, "y": 1864 }, "switch_top1.expert6": { "x": 24, "y": 2014 }, "add@switch_top1#5": { "x": 24, "y": 2164 }, "switch_top1.expert7": { "x": 24, "y": 2314 }, "add@switch_top1#6": { "x": 24, "y": 2464 }, "switch_top1.router.logits": { "x": 24, "y": 214 }, "switch_top1.router.noise_scale": { "x": 24, "y": 364 }, "softplus@switch_top1.router#0": { "x": 24, "y": 514 }, "mul@switch_top1.router#0": { "x": 24, "y": 664 }, "add@switch_top1.router#0": { "x": 24, "y": 814 }, "python@switch_top1.router#0": { "x": 24, "y": 964 }, "python@switch_top1.router#1": { "x": 24, "y": 1114 }, "python@switch_top1.router#2": { "x": 24, "y": 1264 }, "flatten@switch_top1.expert0#0": { "x": 24, "y": 214 }, "python@switch_top1.expert0#0": { "x": 24, "y": 364 }, "python@switch_top1.expert0#1": { "x": 24, "y": 514 }, "switch_top1.expert0.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert0#0": { "x": 24, "y": 814 }, "python@switch_top1.expert0#2": { "x": 24, "y": 964 }, "switch_top1.expert0.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert0.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert0.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert1#0": { "x": 24, "y": 214 }, "python@switch_top1.expert1#0": { "x": 24, "y": 364 }, "python@switch_top1.expert1#1": { "x": 24, "y": 514 }, "switch_top1.expert1.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert1#0": { "x": 24, "y": 814 }, "python@switch_top1.expert1#2": { "x": 24, "y": 964 }, "switch_top1.expert1.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert1.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert1.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert2#0": { "x": 24, "y": 214 }, "python@switch_top1.expert2#0": { "x": 24, "y": 364 }, "python@switch_top1.expert2#1": { "x": 24, "y": 514 }, "switch_top1.expert2.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert2#0": { "x": 24, "y": 814 }, "python@switch_top1.expert2#2": { "x": 24, "y": 964 }, "switch_top1.expert2.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert2.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert2.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert3#0": { "x": 24, "y": 214 }, "python@switch_top1.expert3#0": { "x": 24, "y": 364 }, "python@switch_top1.expert3#1": { "x": 24, "y": 514 }, "switch_top1.expert3.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert3#0": { "x": 24, "y": 814 }, "python@switch_top1.expert3#2": { "x": 24, "y": 964 }, "switch_top1.expert3.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert3.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert3.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert4#0": { "x": 24, "y": 214 }, "python@switch_top1.expert4#0": { "x": 24, "y": 364 }, "python@switch_top1.expert4#1": { "x": 24, "y": 514 }, "switch_top1.expert4.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert4#0": { "x": 24, "y": 814 }, "python@switch_top1.expert4#2": { "x": 24, "y": 964 }, "switch_top1.expert4.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert4.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert4.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert5#0": { "x": 24, "y": 214 }, "python@switch_top1.expert5#0": { "x": 24, "y": 364 }, "python@switch_top1.expert5#1": { "x": 24, "y": 514 }, "switch_top1.expert5.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert5#0": { "x": 24, "y": 814 }, "python@switch_top1.expert5#2": { "x": 24, "y": 964 }, "switch_top1.expert5.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert5.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert5.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert6#0": { "x": 24, "y": 214 }, "python@switch_top1.expert6#0": { "x": 24, "y": 364 }, "python@switch_top1.expert6#1": { "x": 24, "y": 514 }, "switch_top1.expert6.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert6#0": { "x": 24, "y": 814 }, "python@switch_top1.expert6#2": { "x": 24, "y": 964 }, "switch_top1.expert6.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert6.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert6.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@switch_top1.expert7#0": { "x": 24, "y": 214 }, "python@switch_top1.expert7#0": { "x": 24, "y": 364 }, "python@switch_top1.expert7#1": { "x": 24, "y": 514 }, "switch_top1.expert7.expert": { "x": 24, "y": 664 }, "mul@switch_top1.expert7#0": { "x": 24, "y": 814 }, "python@switch_top1.expert7#2": { "x": 24, "y": 964 }, "switch_top1.expert7.expert.gate": { "x": 24, "y": 214 }, "switch_top1.expert7.expert.up": { "x": 24, "y": 364 }, "switch_top1.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@switch_top1.expert7.expert#0": { "x": 24, "y": 664 }, "switch_top1.expert7.expert.down": { "x": 24, "y": 814 }, "mixtral_top2.router": { "x": 24, "y": 214 }, "mixtral_top2.expert0": { "x": 24, "y": 364 }, "mixtral_top2.expert1": { "x": 24, "y": 514 }, "add@mixtral_top2#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2": { "x": 24, "y": 814 }, "add@mixtral_top2#1": { "x": 24, "y": 964 }, "mixtral_top2.expert3": { "x": 24, "y": 1114 }, "add@mixtral_top2#2": { "x": 24, "y": 1264 }, "mixtral_top2.expert4": { "x": 24, "y": 1414 }, "add@mixtral_top2#3": { "x": 24, "y": 1564 }, "mixtral_top2.expert5": { "x": 24, "y": 1714 }, "add@mixtral_top2#4": { "x": 24, "y": 1864 }, "mixtral_top2.expert6": { "x": 24, "y": 2014 }, "add@mixtral_top2#5": { "x": 24, "y": 2164 }, "mixtral_top2.expert7": { "x": 24, "y": 2314 }, "add@mixtral_top2#6": { "x": 24, "y": 2464 }, "mixtral_top2.router.logits": { "x": 24, "y": 214 }, "mixtral_top2.router.noise_scale": { "x": 24, "y": 364 }, "softplus@mixtral_top2.router#0": { "x": 24, "y": 514 }, "mul@mixtral_top2.router#0": { "x": 24, "y": 664 }, "add@mixtral_top2.router#0": { "x": 24, "y": 814 }, "python@mixtral_top2.router#0": { "x": 24, "y": 964 }, "python@mixtral_top2.router#1": { "x": 24, "y": 1114 }, "python@mixtral_top2.router#2": { "x": 24, "y": 1264 }, "flatten@mixtral_top2.expert0#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert0#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert0#1": { "x": 24, "y": 514 }, "mixtral_top2.expert0.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert0#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert0#2": { "x": 24, "y": 964 }, "mixtral_top2.expert0.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert0.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert0.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert0.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert0.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert1#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert1#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert1#1": { "x": 24, "y": 514 }, "mixtral_top2.expert1.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert1#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert1#2": { "x": 24, "y": 964 }, "mixtral_top2.expert1.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert1.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert1.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert1.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert1.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert2#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert2#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert2#1": { "x": 24, "y": 514 }, "mixtral_top2.expert2.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert2#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert2#2": { "x": 24, "y": 964 }, "mixtral_top2.expert2.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert2.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert2.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert2.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert2.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert3#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert3#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert3#1": { "x": 24, "y": 514 }, "mixtral_top2.expert3.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert3#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert3#2": { "x": 24, "y": 964 }, "mixtral_top2.expert3.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert3.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert3.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert3.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert3.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert4#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert4#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert4#1": { "x": 24, "y": 514 }, "mixtral_top2.expert4.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert4#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert4#2": { "x": 24, "y": 964 }, "mixtral_top2.expert4.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert4.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert4.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert4.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert4.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert5#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert5#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert5#1": { "x": 24, "y": 514 }, "mixtral_top2.expert5.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert5#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert5#2": { "x": 24, "y": 964 }, "mixtral_top2.expert5.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert5.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert5.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert5.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert5.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert6#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert6#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert6#1": { "x": 24, "y": 514 }, "mixtral_top2.expert6.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert6#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert6#2": { "x": 24, "y": 964 }, "mixtral_top2.expert6.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert6.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert6.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert6.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert6.expert.down": { "x": 24, "y": 814 }, "flatten@mixtral_top2.expert7#0": { "x": 24, "y": 214 }, "python@mixtral_top2.expert7#0": { "x": 24, "y": 364 }, "python@mixtral_top2.expert7#1": { "x": 24, "y": 514 }, "mixtral_top2.expert7.expert": { "x": 24, "y": 664 }, "mul@mixtral_top2.expert7#0": { "x": 24, "y": 814 }, "python@mixtral_top2.expert7#2": { "x": 24, "y": 964 }, "mixtral_top2.expert7.expert.gate": { "x": 24, "y": 214 }, "mixtral_top2.expert7.expert.up": { "x": 24, "y": 364 }, "mixtral_top2.expert7.expert.silu": { "x": 24, "y": 514 }, "mul@mixtral_top2.expert7.expert#0": { "x": 24, "y": 664 }, "mixtral_top2.expert7.expert.down": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "mixtral_top2" ], "selectedNodeIds": [ "mixtral_top2.router" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "mixtral_top2.router" }, "snapshotId": "snapshot-82f8c493b242ec59", "exportedAt": "2026-09-17T06:02:58.595933+00:00", "modelName": "SparseRoutingComparison", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.router.logits\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.router.noise_scale\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert3\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert4\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert5\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert6\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mixtral_top2.expert7\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router.logits\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mixtral_top2.router.noise_scale\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"softplus@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#1\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.router#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"flatten@mixtral_top2.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert0#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert1#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert1#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert2#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert2#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert3#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert3#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert4#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert4#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert5#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert5#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert6#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert6#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert7#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@mixtral_top2.router#2\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"python@mixtral_top2.expert7#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"softplus@mixtral_top2.router#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[\"{\\\"id\\\":\\\"add@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.add\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.router.logits\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"mixtral_top2.router.noise_scale\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"bias\\\":false,\\\"in_features\\\":8,\\\"out_features\\\":8},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"mul@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.mul\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.router#1\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"python@mixtral_top2.router#2\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Python\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\",\"{\\\"id\\\":\\\"softplus@mixtral_top2.router#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Softplus\\\",\\\"params\\\":{},\\\"parent\\\":\\\"mixtral_top2.router\\\"}\"],\"node\":{\"id\":\"mixtral_top2.router\",\"kind\":\"module\",\"op\":\"Module\",\"params\":{\"arg_0\":2,\"arg_1\":true,\"arg_2\":false},\"parent\":\"mixtral_top2\"}}" } ] } ], "openingViewId": "route" } }