{ "format": "tensorviz.graph", "version": 3, "exportedAt": "2026-09-17T07:00:38.449307+00:00", "status": "static", "warningCount": 0, "document": { "schemaVersion": 2, "semantic": { "graph_id": "snapshot", "name": "PromptableMaskDecoder", "revision": 0, "nodes": [ { "node_id": "input", "kind": "input", "op": "Input", "params": { "shape": "1,3,8" }, "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": "flatten@PromptableMaskDecoder#0", "kind": "layer", "op": "Flatten", "params": { "start_dim": 2 }, "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" } ] } } }, { "node_id": "prompt_encoder", "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" } ] }, "moduleInstanceId": "prompt_encoder", "moduleTemplateId": "model::PromptEncoder", "isGroup": true, "moduleMembers": [ "prompt_encoder.position", "prompt_encoder.type_embedding", "add@prompt_encoder#0" ], "moduleLayerCount": 3 } }, { "node_id": "python@PromptableMaskDecoder#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": [] } } }, { "node_id": "output_tokens", "kind": "layer", "op": "Parameter", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "parameter", "description": "", "category": "misc", "icon": "Box", "parameters": [], "name": "Parameter", "displayName": "Parameter" } } }, { "node_id": "cat@PromptableMaskDecoder#0", "kind": "layer", "op": "torch.cat", "params": { "dim": 1 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "cat", "name": "torch.cat", "displayName": "torch.cat", "description": "PyTorch cat operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] } } }, { "node_id": "no_mask", "kind": "layer", "op": "Parameter", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "parameter", "description": "", "category": "misc", "icon": "Box", "parameters": [], "name": "Parameter", "displayName": "Parameter" } } }, { "node_id": "add@PromptableMaskDecoder#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" } ] } } }, { "node_id": "decoder", "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" } ] }, "moduleInstanceId": "decoder", "moduleTemplateId": "model::TwoWayDecoder", "isGroup": true, "moduleMembers": [ "decoder.first", "decoder.second", "add@decoder#0", "add@decoder#1", "decoder.final_attention", "add@decoder#2", "decoder.final_norm" ], "moduleLayerCount": 72 } }, { "node_id": "upscale", "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" } ] }, "moduleInstanceId": "upscale", "moduleTemplateId": "model::Upscale", "isGroup": true, "moduleMembers": [ "upscale.first", "permute@upscale#0", "upscale.norm", "permute@upscale#1", "gelu@upscale#0", "upscale.second", "gelu@upscale#1" ], "moduleLayerCount": 7 } }, { "node_id": "masks", "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" } ] }, "moduleInstanceId": "masks", "moduleTemplateId": "model::MaskHeads", "isGroup": true, "moduleMembers": [ "flatten@masks#0", "masks.mask0", "masks.mask1", "masks.mask2", "stack@masks#0", "python@masks#0" ], "moduleLayerCount": 18 } }, { "node_id": "flatten@PromptableMaskDecoder#1", "kind": "layer", "op": "Flatten", "params": { "start_dim": 1 }, "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" } ] } } }, { "node_id": "quality", "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" } ] }, "moduleInstanceId": "quality", "moduleTemplateId": "model::QualityHead", "isGroup": true, "moduleMembers": [ "quality.first", "relu@quality#0", "quality.second", "relu@quality#1", "quality.scores" ], "moduleLayerCount": 5 } }, { "node_id": "cat@PromptableMaskDecoder#1", "kind": "layer", "op": "torch.cat", "params": { "dim": -1 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "cat", "name": "torch.cat", "displayName": "torch.cat", "description": "PyTorch cat 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": "prompt_encoder.position", "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": "prompt_encoder", "moduleInstanceId": "prompt_encoder.position", "moduleTemplateId": "model::PositionEncoding", "isGroup": true, "moduleMembers": [ "cat@prompt_encoder.position#0" ], "moduleLayerCount": 1 } }, { "node_id": "prompt_encoder.type_embedding", "kind": "layer", "op": "Embedding", "params": { "num_embeddings": 4, "embedding_dim": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "embedding", "name": "Embedding", "displayName": "Embedding", "description": "Lookup table for dense vector embeddings", "icon": "Hash", "category": "embedding", "parameters": [ { "name": "num_embeddings", "type": "number", "defaultValue": 10000, "required": true, "description": "Size of the dictionary of embeddings", "min": 1 }, { "name": "embedding_dim", "type": "number", "defaultValue": 128, "required": true, "description": "Size of each embedding vector", "min": 1 }, { "name": "padding_idx", "type": "number", "defaultValue": 0, "required": false, "description": "Pad output with zeros at this index", "min": 0 }, { "name": "max_norm", "type": "number", "defaultValue": 0, "required": false, "description": "If > 0, re-normalizes embeddings with norm > max_norm", "min": 0 }, { "name": "sparse", "type": "boolean", "defaultValue": false, "required": false, "description": "If True, gradient w.r.t. weight is sparse" } ] }, "parentId": "prompt_encoder" } }, { "node_id": "add@prompt_encoder#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": "prompt_encoder" } }, { "node_id": "cat@prompt_encoder.position#0", "kind": "layer", "op": "torch.cat", "params": { "dim": -1 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "cat", "name": "torch.cat", "displayName": "torch.cat", "description": "PyTorch cat operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "prompt_encoder.position" } }, { "node_id": "decoder.first", "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": "decoder", "moduleInstanceId": "decoder.first", "moduleTemplateId": "model::TwoWayBlock", "isGroup": true, "moduleMembers": [ "decoder.first.self_attention", "add@decoder.first#0", "decoder.first.self_norm", "add@decoder.first#1", "decoder.first.token_to_image", "add@decoder.first#2", "decoder.first.token_norm", "decoder.first.up", "decoder.first.relu", "decoder.first.down", "add@decoder.first#3", "decoder.first.mlp_norm", "add@decoder.first#4", "decoder.first.image_to_token", "add@decoder.first#5", "decoder.first.image_norm" ], "moduleLayerCount": 31 } }, { "node_id": "decoder.second", "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": "decoder", "moduleInstanceId": "decoder.second", "moduleTemplateId": "model::TwoWayBlock", "isGroup": true, "moduleMembers": [ "decoder.second.self_attention", "add@decoder.second#0", "decoder.second.self_norm", "add@decoder.second#1", "decoder.second.token_to_image", "add@decoder.second#2", "decoder.second.token_norm", "decoder.second.up", "decoder.second.relu", "decoder.second.down", "add@decoder.second#3", "decoder.second.mlp_norm", "add@decoder.second#4", "decoder.second.image_to_token", "add@decoder.second#5", "decoder.second.image_norm" ], "moduleLayerCount": 31 } }, { "node_id": "add@decoder#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": "decoder" } }, { "node_id": "add@decoder#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": "decoder" } }, { "node_id": "decoder.final_attention", "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": "decoder", "moduleInstanceId": "decoder.final_attention", "moduleTemplateId": "model::Attention", "isGroup": true, "moduleMembers": [ "decoder.final_attention.query", "decoder.final_attention.key", "div@decoder.final_attention#0", "decoder.final_attention.value", "decoder.final_attention.softmax", "decoder.final_attention.project" ], "moduleLayerCount": 6 } }, { "node_id": "add@decoder#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": "decoder" } }, { "node_id": "decoder.final_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder" } }, { "node_id": "decoder.first.self_attention", "kind": "module", "op": "Module", "params": { "arg_0": 8 }, "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": "decoder.first", "moduleInstanceId": "decoder.first.self_attention", "moduleTemplateId": "model::Attention", "isGroup": true, "moduleMembers": [ "decoder.first.self_attention.query", "decoder.first.self_attention.key", "div@decoder.first.self_attention#0", "decoder.first.self_attention.value", "decoder.first.self_attention.softmax", "decoder.first.self_attention.project" ], "moduleLayerCount": 6 } }, { "node_id": "add@decoder.first#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": "decoder.first" } }, { "node_id": "decoder.first.self_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.first" } }, { "node_id": "add@decoder.first#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": "decoder.first" } }, { "node_id": "decoder.first.token_to_image", "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": "decoder.first", "moduleInstanceId": "decoder.first.token_to_image", "moduleTemplateId": "model::Attention", "isGroup": true, "moduleMembers": [ "decoder.first.token_to_image.query", "decoder.first.token_to_image.key", "div@decoder.first.token_to_image#0", "decoder.first.token_to_image.value", "decoder.first.token_to_image.softmax", "decoder.first.token_to_image.project" ], "moduleLayerCount": 6 } }, { "node_id": "add@decoder.first#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": "decoder.first" } }, { "node_id": "decoder.first.token_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.first" } }, { "node_id": "decoder.first.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 32 }, "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": "decoder.first" } }, { "node_id": "decoder.first.relu", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "decoder.first" } }, { "node_id": "decoder.first.down", "kind": "layer", "op": "Linear", "params": { "in_features": 32, "out_features": 8 }, "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": "decoder.first" } }, { "node_id": "add@decoder.first#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": "decoder.first" } }, { "node_id": "decoder.first.mlp_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.first" } }, { "node_id": "add@decoder.first#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": "decoder.first" } }, { "node_id": "decoder.first.image_to_token", "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": "decoder.first", "moduleInstanceId": "decoder.first.image_to_token", "moduleTemplateId": "model::Attention", "isGroup": true, "moduleMembers": [ "decoder.first.image_to_token.query", "decoder.first.image_to_token.key", "div@decoder.first.image_to_token#0", "decoder.first.image_to_token.value", "decoder.first.image_to_token.softmax", "decoder.first.image_to_token.project" ], "moduleLayerCount": 6 } }, { "node_id": "add@decoder.first#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": "decoder.first" } }, { "node_id": "decoder.first.image_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.first" } }, { "node_id": "decoder.first.self_attention.query", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.self_attention" } }, { "node_id": "decoder.first.self_attention.key", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.self_attention" } }, { "node_id": "div@decoder.first.self_attention#0", "kind": "layer", "op": "torch.div", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "div", "name": "torch.div", "displayName": "torch.div", "description": "PyTorch div operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "decoder.first.self_attention" } }, { "node_id": "decoder.first.self_attention.value", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.self_attention" } }, { "node_id": "decoder.first.self_attention.softmax", "kind": "layer", "op": "Softmax", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softmax", "name": "Softmax", "displayName": "Softmax", "description": "Softmax over a dimension", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "dim", "type": "number", "defaultValue": 1, "required": true, "description": "Dimension along which Softmax is computed" } ] }, "parentId": "decoder.first.self_attention" } }, { "node_id": "decoder.first.self_attention.project", "kind": "layer", "op": "Linear", "params": { "in_features": 4, "out_features": 8 }, "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": "decoder.first.self_attention" } }, { "node_id": "decoder.first.token_to_image.query", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.token_to_image" } }, { "node_id": "decoder.first.token_to_image.key", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.token_to_image" } }, { "node_id": "div@decoder.first.token_to_image#0", "kind": "layer", "op": "torch.div", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "div", "name": "torch.div", "displayName": "torch.div", "description": "PyTorch div operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "decoder.first.token_to_image" } }, { "node_id": "decoder.first.token_to_image.value", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.token_to_image" } }, { "node_id": "decoder.first.token_to_image.softmax", "kind": "layer", "op": "Softmax", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softmax", "name": "Softmax", "displayName": "Softmax", "description": "Softmax over a dimension", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "dim", "type": "number", "defaultValue": 1, "required": true, "description": "Dimension along which Softmax is computed" } ] }, "parentId": "decoder.first.token_to_image" } }, { "node_id": "decoder.first.token_to_image.project", "kind": "layer", "op": "Linear", "params": { "in_features": 4, "out_features": 8 }, "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": "decoder.first.token_to_image" } }, { "node_id": "decoder.first.image_to_token.query", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.image_to_token" } }, { "node_id": "decoder.first.image_to_token.key", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.image_to_token" } }, { "node_id": "div@decoder.first.image_to_token#0", "kind": "layer", "op": "torch.div", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "div", "name": "torch.div", "displayName": "torch.div", "description": "PyTorch div operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "decoder.first.image_to_token" } }, { "node_id": "decoder.first.image_to_token.value", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.first.image_to_token" } }, { "node_id": "decoder.first.image_to_token.softmax", "kind": "layer", "op": "Softmax", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softmax", "name": "Softmax", "displayName": "Softmax", "description": "Softmax over a dimension", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "dim", "type": "number", "defaultValue": 1, "required": true, "description": "Dimension along which Softmax is computed" } ] }, "parentId": "decoder.first.image_to_token" } }, { "node_id": "decoder.first.image_to_token.project", "kind": "layer", "op": "Linear", "params": { "in_features": 4, "out_features": 8 }, "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": "decoder.first.image_to_token" } }, { "node_id": "decoder.second.self_attention", "kind": "module", "op": "Module", "params": { "arg_0": 8 }, "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": "decoder.second", "moduleInstanceId": "decoder.second.self_attention", "moduleTemplateId": "model::Attention", "isGroup": true, "moduleMembers": [ "decoder.second.self_attention.query", "decoder.second.self_attention.key", "div@decoder.second.self_attention#0", "decoder.second.self_attention.value", "decoder.second.self_attention.softmax", "decoder.second.self_attention.project" ], "moduleLayerCount": 6 } }, { "node_id": "add@decoder.second#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": "decoder.second" } }, { "node_id": "decoder.second.self_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.second" } }, { "node_id": "add@decoder.second#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": "decoder.second" } }, { "node_id": "decoder.second.token_to_image", "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": "decoder.second", "moduleInstanceId": "decoder.second.token_to_image", "moduleTemplateId": "model::Attention", "isGroup": true, "moduleMembers": [ "decoder.second.token_to_image.query", "decoder.second.token_to_image.key", "div@decoder.second.token_to_image#0", "decoder.second.token_to_image.value", "decoder.second.token_to_image.softmax", "decoder.second.token_to_image.project" ], "moduleLayerCount": 6 } }, { "node_id": "add@decoder.second#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": "decoder.second" } }, { "node_id": "decoder.second.token_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.second" } }, { "node_id": "decoder.second.up", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 32 }, "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": "decoder.second" } }, { "node_id": "decoder.second.relu", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "decoder.second" } }, { "node_id": "decoder.second.down", "kind": "layer", "op": "Linear", "params": { "in_features": 32, "out_features": 8 }, "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": "decoder.second" } }, { "node_id": "add@decoder.second#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": "decoder.second" } }, { "node_id": "decoder.second.mlp_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.second" } }, { "node_id": "add@decoder.second#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": "decoder.second" } }, { "node_id": "decoder.second.image_to_token", "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": "decoder.second", "moduleInstanceId": "decoder.second.image_to_token", "moduleTemplateId": "model::Attention", "isGroup": true, "moduleMembers": [ "decoder.second.image_to_token.query", "decoder.second.image_to_token.key", "div@decoder.second.image_to_token#0", "decoder.second.image_to_token.value", "decoder.second.image_to_token.softmax", "decoder.second.image_to_token.project" ], "moduleLayerCount": 6 } }, { "node_id": "add@decoder.second#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": "decoder.second" } }, { "node_id": "decoder.second.image_norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 8 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "decoder.second" } }, { "node_id": "decoder.second.self_attention.query", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.self_attention" } }, { "node_id": "decoder.second.self_attention.key", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.self_attention" } }, { "node_id": "div@decoder.second.self_attention#0", "kind": "layer", "op": "torch.div", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "div", "name": "torch.div", "displayName": "torch.div", "description": "PyTorch div operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "decoder.second.self_attention" } }, { "node_id": "decoder.second.self_attention.value", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.self_attention" } }, { "node_id": "decoder.second.self_attention.softmax", "kind": "layer", "op": "Softmax", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softmax", "name": "Softmax", "displayName": "Softmax", "description": "Softmax over a dimension", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "dim", "type": "number", "defaultValue": 1, "required": true, "description": "Dimension along which Softmax is computed" } ] }, "parentId": "decoder.second.self_attention" } }, { "node_id": "decoder.second.self_attention.project", "kind": "layer", "op": "Linear", "params": { "in_features": 4, "out_features": 8 }, "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": "decoder.second.self_attention" } }, { "node_id": "decoder.second.token_to_image.query", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.token_to_image" } }, { "node_id": "decoder.second.token_to_image.key", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.token_to_image" } }, { "node_id": "div@decoder.second.token_to_image#0", "kind": "layer", "op": "torch.div", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "div", "name": "torch.div", "displayName": "torch.div", "description": "PyTorch div operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "decoder.second.token_to_image" } }, { "node_id": "decoder.second.token_to_image.value", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.token_to_image" } }, { "node_id": "decoder.second.token_to_image.softmax", "kind": "layer", "op": "Softmax", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softmax", "name": "Softmax", "displayName": "Softmax", "description": "Softmax over a dimension", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "dim", "type": "number", "defaultValue": 1, "required": true, "description": "Dimension along which Softmax is computed" } ] }, "parentId": "decoder.second.token_to_image" } }, { "node_id": "decoder.second.token_to_image.project", "kind": "layer", "op": "Linear", "params": { "in_features": 4, "out_features": 8 }, "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": "decoder.second.token_to_image" } }, { "node_id": "decoder.second.image_to_token.query", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.image_to_token" } }, { "node_id": "decoder.second.image_to_token.key", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.image_to_token" } }, { "node_id": "div@decoder.second.image_to_token#0", "kind": "layer", "op": "torch.div", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "div", "name": "torch.div", "displayName": "torch.div", "description": "PyTorch div operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "decoder.second.image_to_token" } }, { "node_id": "decoder.second.image_to_token.value", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.second.image_to_token" } }, { "node_id": "decoder.second.image_to_token.softmax", "kind": "layer", "op": "Softmax", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softmax", "name": "Softmax", "displayName": "Softmax", "description": "Softmax over a dimension", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "dim", "type": "number", "defaultValue": 1, "required": true, "description": "Dimension along which Softmax is computed" } ] }, "parentId": "decoder.second.image_to_token" } }, { "node_id": "decoder.second.image_to_token.project", "kind": "layer", "op": "Linear", "params": { "in_features": 4, "out_features": 8 }, "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": "decoder.second.image_to_token" } }, { "node_id": "decoder.final_attention.query", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.final_attention" } }, { "node_id": "decoder.final_attention.key", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.final_attention" } }, { "node_id": "div@decoder.final_attention#0", "kind": "layer", "op": "torch.div", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "div", "name": "torch.div", "displayName": "torch.div", "description": "PyTorch div operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dim", "type": "string", "defaultValue": "", "required": false, "description": "Dimension(s) to apply the operation on" } ] }, "parentId": "decoder.final_attention" } }, { "node_id": "decoder.final_attention.value", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 4 }, "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": "decoder.final_attention" } }, { "node_id": "decoder.final_attention.softmax", "kind": "layer", "op": "Softmax", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "softmax", "name": "Softmax", "displayName": "Softmax", "description": "Softmax over a dimension", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "dim", "type": "number", "defaultValue": 1, "required": true, "description": "Dimension along which Softmax is computed" } ] }, "parentId": "decoder.final_attention" } }, { "node_id": "decoder.final_attention.project", "kind": "layer", "op": "Linear", "params": { "in_features": 4, "out_features": 8 }, "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": "decoder.final_attention" } }, { "node_id": "upscale.first", "kind": "layer", "op": "ConvTranspose2d", "params": { "in_channels": 8, "out_channels": 4, "kernel_size": 2, "stride": 2 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "convtranspose2d", "name": "ConvTranspose2d", "displayName": "ConvTranspose2D", "description": "2D Transposed Convolutional Layer", "icon": "ChevronsRight", "category": "conv", "parameters": [ { "name": "in_channels", "type": "number", "defaultValue": 64, "required": true, "description": "Number of input channels", "min": 1 }, { "name": "out_channels", "type": "number", "defaultValue": 32, "required": true, "description": "Number of output channels", "min": 1 }, { "name": "kernel_size", "type": "number", "defaultValue": 3, "required": true, "description": "Size of the convolving kernel", "min": 1 }, { "name": "stride", "type": "number", "defaultValue": 1, "required": false, "description": "Stride of the convolution", "min": 1 }, { "name": "padding", "type": "number", "defaultValue": 0, "required": false, "description": "Zero-padding added to input", "min": 0 }, { "name": "output_padding", "type": "number", "defaultValue": 0, "required": false, "description": "Additional size added to one side of the output shape", "min": 0 }, { "name": "groups", "type": "number", "defaultValue": 1, "required": false, "description": "Number of blocked connections", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, adds a learnable bias" } ] }, "parentId": "upscale" } }, { "node_id": "permute@upscale#0", "kind": "layer", "op": "torch.permute", "params": { "dims": [ 0, 2, 3, 1 ] }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "permute", "name": "torch.permute", "displayName": "torch.permute", "description": "PyTorch permute operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dims", "type": "string", "defaultValue": "", "required": false, "description": "The new dimension order — one entry per dimension, e.g. 0,2,3,1" } ] }, "parentId": "upscale" } }, { "node_id": "upscale.norm", "kind": "layer", "op": "LayerNorm", "params": { "normalized_shape": 4 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "layernorm", "name": "LayerNorm", "displayName": "LayerNorm", "description": "Layer Normalization", "icon": "Layers", "category": "norm", "parameters": [ { "name": "normalized_shape", "type": "string", "defaultValue": "512", "required": true, "description": "Input shape for normalization (comma-separated for multi-dim, e.g., 512)" }, { "name": "eps", "type": "number", "defaultValue": 0.00001, "required": false, "description": "Value for numerical stability", "min": 0 }, { "name": "elementwise_affine", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, learnable affine parameters" } ] }, "parentId": "upscale" } }, { "node_id": "permute@upscale#1", "kind": "layer", "op": "torch.permute", "params": { "dims": [ 0, 3, 1, 2 ] }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "permute", "name": "torch.permute", "displayName": "torch.permute", "description": "PyTorch permute operation", "icon": "Activity", "category": "tensor", "parameters": [ { "name": "dims", "type": "string", "defaultValue": "", "required": false, "description": "The new dimension order — one entry per dimension, e.g. 0,2,3,1" } ] }, "parentId": "upscale" } }, { "node_id": "gelu@upscale#0", "kind": "layer", "op": "GELU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "gelu", "name": "GELU", "displayName": "GELU", "description": "Gaussian Error Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "approximate", "type": "select", "defaultValue": "none", "required": false, "description": "The gelu approximation algorithm to use: 'none' or 'tanh'", "options": [ "none", "tanh" ] } ] }, "parentId": "upscale" } }, { "node_id": "upscale.second", "kind": "layer", "op": "ConvTranspose2d", "params": { "in_channels": 4, "out_channels": 2, "kernel_size": 2, "stride": 2 }, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "convtranspose2d", "name": "ConvTranspose2d", "displayName": "ConvTranspose2D", "description": "2D Transposed Convolutional Layer", "icon": "ChevronsRight", "category": "conv", "parameters": [ { "name": "in_channels", "type": "number", "defaultValue": 64, "required": true, "description": "Number of input channels", "min": 1 }, { "name": "out_channels", "type": "number", "defaultValue": 32, "required": true, "description": "Number of output channels", "min": 1 }, { "name": "kernel_size", "type": "number", "defaultValue": 3, "required": true, "description": "Size of the convolving kernel", "min": 1 }, { "name": "stride", "type": "number", "defaultValue": 1, "required": false, "description": "Stride of the convolution", "min": 1 }, { "name": "padding", "type": "number", "defaultValue": 0, "required": false, "description": "Zero-padding added to input", "min": 0 }, { "name": "output_padding", "type": "number", "defaultValue": 0, "required": false, "description": "Additional size added to one side of the output shape", "min": 0 }, { "name": "groups", "type": "number", "defaultValue": 1, "required": false, "description": "Number of blocked connections", "min": 1 }, { "name": "bias", "type": "boolean", "defaultValue": true, "required": false, "description": "If True, adds a learnable bias" } ] }, "parentId": "upscale" } }, { "node_id": "gelu@upscale#1", "kind": "layer", "op": "GELU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "gelu", "name": "GELU", "displayName": "GELU", "description": "Gaussian Error Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "approximate", "type": "select", "defaultValue": "none", "required": false, "description": "The gelu approximation algorithm to use: 'none' or 'tanh'", "options": [ "none", "tanh" ] } ] }, "parentId": "upscale" } }, { "node_id": "flatten@masks#0", "kind": "layer", "op": "Flatten", "params": { "start_dim": 2 }, "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": "masks" } }, { "node_id": "masks.mask0", "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": "masks", "moduleInstanceId": "masks.mask0", "moduleTemplateId": "model::Hypernetwork", "isGroup": true, "moduleMembers": [ "masks.mask0.first", "relu@masks.mask0#0", "masks.mask0.second", "relu@masks.mask0#1", "masks.mask0.project" ], "moduleLayerCount": 5 } }, { "node_id": "masks.mask1", "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": "masks", "moduleInstanceId": "masks.mask1", "moduleTemplateId": "model::Hypernetwork", "isGroup": true, "moduleMembers": [ "masks.mask1.first", "relu@masks.mask1#0", "masks.mask1.second", "relu@masks.mask1#1", "masks.mask1.project" ], "moduleLayerCount": 5 } }, { "node_id": "masks.mask2", "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": "masks", "moduleInstanceId": "masks.mask2", "moduleTemplateId": "model::Hypernetwork", "isGroup": true, "moduleMembers": [ "masks.mask2.first", "relu@masks.mask2#0", "masks.mask2.second", "relu@masks.mask2#1", "masks.mask2.project" ], "moduleLayerCount": 5 } }, { "node_id": "stack@masks#0", "kind": "layer", "op": "torch.stack", "params": { "dim": 1 }, "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" } ] }, "parentId": "masks" } }, { "node_id": "python@masks#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": "masks" } }, { "node_id": "masks.mask0.first", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "masks.mask0" } }, { "node_id": "relu@masks.mask0#0", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "masks.mask0" } }, { "node_id": "masks.mask0.second", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "masks.mask0" } }, { "node_id": "relu@masks.mask0#1", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "masks.mask0" } }, { "node_id": "masks.mask0.project", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 2 }, "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": "masks.mask0" } }, { "node_id": "masks.mask1.first", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "masks.mask1" } }, { "node_id": "relu@masks.mask1#0", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "masks.mask1" } }, { "node_id": "masks.mask1.second", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "masks.mask1" } }, { "node_id": "relu@masks.mask1#1", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "masks.mask1" } }, { "node_id": "masks.mask1.project", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 2 }, "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": "masks.mask1" } }, { "node_id": "masks.mask2.first", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "masks.mask2" } }, { "node_id": "relu@masks.mask2#0", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "masks.mask2" } }, { "node_id": "masks.mask2.second", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "masks.mask2" } }, { "node_id": "relu@masks.mask2#1", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "masks.mask2" } }, { "node_id": "masks.mask2.project", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 2 }, "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": "masks.mask2" } }, { "node_id": "quality.first", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "quality" } }, { "node_id": "relu@quality#0", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "quality" } }, { "node_id": "quality.second", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 8 }, "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": "quality" } }, { "node_id": "relu@quality#1", "kind": "layer", "op": "ReLU", "params": {}, "in_ports": [ "x" ], "out_ports": [ "y" ], "metadata": { "layerType": { "id": "relu", "name": "ReLU", "displayName": "ReLU", "description": "Rectified Linear Unit", "icon": "TrendingUp", "category": "activation", "parameters": [ { "name": "inplace", "type": "boolean", "defaultValue": false, "required": false, "description": "Can optionally do the operation in-place" } ] }, "parentId": "quality" } }, { "node_id": "quality.scores", "kind": "layer", "op": "Linear", "params": { "in_features": 8, "out_features": 3 }, "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": "quality" } } ], "edges": [ { "edge_id": "edge-1", "source_node_id": "cat@prompt_encoder.position#0", "target_node_id": "add@prompt_encoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-2", "source_node_id": "prompt_encoder.position", "target_node_id": "add@prompt_encoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-3", "source_node_id": "prompt_encoder.type_embedding", "target_node_id": "add@prompt_encoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-4", "source_node_id": "decoder.first.self_attention.query", "target_node_id": "div@decoder.first.self_attention#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-5", "source_node_id": "decoder.first.self_attention.key", "target_node_id": "div@decoder.first.self_attention#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-6", "source_node_id": "div@decoder.first.self_attention#0", "target_node_id": "decoder.first.self_attention.softmax", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-7", "source_node_id": "decoder.first.self_attention.softmax", "target_node_id": "decoder.first.self_attention.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-8", "source_node_id": "decoder.first.self_attention.value", "target_node_id": "decoder.first.self_attention.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-9", "source_node_id": "decoder.first.token_to_image.query", "target_node_id": "div@decoder.first.token_to_image#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-10", "source_node_id": "decoder.first.token_to_image.key", "target_node_id": "div@decoder.first.token_to_image#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-11", "source_node_id": "div@decoder.first.token_to_image#0", "target_node_id": "decoder.first.token_to_image.softmax", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-12", "source_node_id": "decoder.first.token_to_image.softmax", "target_node_id": "decoder.first.token_to_image.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-13", "source_node_id": "decoder.first.token_to_image.value", "target_node_id": "decoder.first.token_to_image.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-14", "source_node_id": "decoder.first.image_to_token.query", "target_node_id": "div@decoder.first.image_to_token#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-15", "source_node_id": "decoder.first.image_to_token.key", "target_node_id": "div@decoder.first.image_to_token#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-16", "source_node_id": "div@decoder.first.image_to_token#0", "target_node_id": "decoder.first.image_to_token.softmax", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-17", "source_node_id": "decoder.first.image_to_token.softmax", "target_node_id": "decoder.first.image_to_token.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-18", "source_node_id": "decoder.first.image_to_token.value", "target_node_id": "decoder.first.image_to_token.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-19", "source_node_id": "decoder.first.self_attention.project", "target_node_id": "add@decoder.first#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-20", "source_node_id": "decoder.first.self_attention", "target_node_id": "add@decoder.first#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-21", "source_node_id": "add@decoder.first#0", "target_node_id": "decoder.first.self_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-22", "source_node_id": "decoder.first.self_norm", "target_node_id": "add@decoder.first#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-23", "source_node_id": "add@decoder.first#1", "target_node_id": "decoder.first.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-24", "source_node_id": "add@decoder.first#1", "target_node_id": "decoder.first.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-25", "source_node_id": "add@decoder.first#1", "target_node_id": "decoder.first.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-26", "source_node_id": "add@decoder.first#1", "target_node_id": "decoder.first.token_to_image", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-27", "source_node_id": "decoder.first.self_norm", "target_node_id": "add@decoder.first#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-28", "source_node_id": "decoder.first.token_to_image.project", "target_node_id": "add@decoder.first#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-29", "source_node_id": "decoder.first.token_to_image", "target_node_id": "add@decoder.first#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-30", "source_node_id": "add@decoder.first#2", "target_node_id": "decoder.first.token_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-31", "source_node_id": "decoder.first.token_norm", "target_node_id": "decoder.first.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-32", "source_node_id": "decoder.first.up", "target_node_id": "decoder.first.relu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-33", "source_node_id": "decoder.first.relu", "target_node_id": "decoder.first.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-34", "source_node_id": "decoder.first.token_norm", "target_node_id": "add@decoder.first#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-35", "source_node_id": "decoder.first.down", "target_node_id": "add@decoder.first#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-36", "source_node_id": "add@decoder.first#3", "target_node_id": "decoder.first.mlp_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-37", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "add@decoder.first#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-38", "source_node_id": "add@decoder.first#4", "target_node_id": "decoder.first.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-39", "source_node_id": "add@decoder.first#4", "target_node_id": "decoder.first.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-40", "source_node_id": "add@decoder.first#4", "target_node_id": "decoder.first.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-41", "source_node_id": "add@decoder.first#4", "target_node_id": "decoder.first.image_to_token", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-42", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.first.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-43", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.first.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-44", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.first.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-45", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.first.image_to_token", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-46", "source_node_id": "decoder.first.image_to_token.project", "target_node_id": "add@decoder.first#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-47", "source_node_id": "decoder.first.image_to_token", "target_node_id": "add@decoder.first#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-48", "source_node_id": "add@decoder.first#5", "target_node_id": "decoder.first.image_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-49", "source_node_id": "decoder.second.self_attention.query", "target_node_id": "div@decoder.second.self_attention#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-50", "source_node_id": "decoder.second.self_attention.key", "target_node_id": "div@decoder.second.self_attention#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-51", "source_node_id": "div@decoder.second.self_attention#0", "target_node_id": "decoder.second.self_attention.softmax", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-52", "source_node_id": "decoder.second.self_attention.softmax", "target_node_id": "decoder.second.self_attention.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-53", "source_node_id": "decoder.second.self_attention.value", "target_node_id": "decoder.second.self_attention.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-54", "source_node_id": "decoder.second.token_to_image.query", "target_node_id": "div@decoder.second.token_to_image#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-55", "source_node_id": "decoder.second.token_to_image.key", "target_node_id": "div@decoder.second.token_to_image#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-56", "source_node_id": "div@decoder.second.token_to_image#0", "target_node_id": "decoder.second.token_to_image.softmax", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-57", "source_node_id": "decoder.second.token_to_image.softmax", "target_node_id": "decoder.second.token_to_image.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-58", "source_node_id": "decoder.second.token_to_image.value", "target_node_id": "decoder.second.token_to_image.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-59", "source_node_id": "decoder.second.image_to_token.query", "target_node_id": "div@decoder.second.image_to_token#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-60", "source_node_id": "decoder.second.image_to_token.key", "target_node_id": "div@decoder.second.image_to_token#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-61", "source_node_id": "div@decoder.second.image_to_token#0", "target_node_id": "decoder.second.image_to_token.softmax", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-62", "source_node_id": "decoder.second.image_to_token.softmax", "target_node_id": "decoder.second.image_to_token.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-63", "source_node_id": "decoder.second.image_to_token.value", "target_node_id": "decoder.second.image_to_token.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-64", "source_node_id": "decoder.second.self_attention.project", "target_node_id": "add@decoder.second#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-65", "source_node_id": "decoder.second.self_attention", "target_node_id": "add@decoder.second#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-66", "source_node_id": "add@decoder.second#0", "target_node_id": "decoder.second.self_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-67", "source_node_id": "decoder.second.self_norm", "target_node_id": "add@decoder.second#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-68", "source_node_id": "add@decoder.second#1", "target_node_id": "decoder.second.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-69", "source_node_id": "add@decoder.second#1", "target_node_id": "decoder.second.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-70", "source_node_id": "add@decoder.second#1", "target_node_id": "decoder.second.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-71", "source_node_id": "add@decoder.second#1", "target_node_id": "decoder.second.token_to_image", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-72", "source_node_id": "decoder.second.self_norm", "target_node_id": "add@decoder.second#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-73", "source_node_id": "decoder.second.token_to_image.project", "target_node_id": "add@decoder.second#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-74", "source_node_id": "decoder.second.token_to_image", "target_node_id": "add@decoder.second#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-75", "source_node_id": "add@decoder.second#2", "target_node_id": "decoder.second.token_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-76", "source_node_id": "decoder.second.token_norm", "target_node_id": "decoder.second.up", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-77", "source_node_id": "decoder.second.up", "target_node_id": "decoder.second.relu", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-78", "source_node_id": "decoder.second.relu", "target_node_id": "decoder.second.down", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-79", "source_node_id": "decoder.second.token_norm", "target_node_id": "add@decoder.second#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-80", "source_node_id": "decoder.second.down", "target_node_id": "add@decoder.second#3", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-81", "source_node_id": "add@decoder.second#3", "target_node_id": "decoder.second.mlp_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-82", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "add@decoder.second#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-83", "source_node_id": "add@decoder.second#4", "target_node_id": "decoder.second.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-84", "source_node_id": "add@decoder.second#4", "target_node_id": "decoder.second.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-85", "source_node_id": "add@decoder.second#4", "target_node_id": "decoder.second.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-86", "source_node_id": "add@decoder.second#4", "target_node_id": "decoder.second.image_to_token", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-87", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "decoder.second.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-88", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "decoder.second.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-89", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "decoder.second.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-90", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "decoder.second.image_to_token", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-91", "source_node_id": "decoder.second.image_to_token.project", "target_node_id": "add@decoder.second#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-92", "source_node_id": "decoder.second.image_to_token", "target_node_id": "add@decoder.second#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-93", "source_node_id": "add@decoder.second#5", "target_node_id": "decoder.second.image_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-94", "source_node_id": "decoder.final_attention.query", "target_node_id": "div@decoder.final_attention#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-95", "source_node_id": "decoder.final_attention.key", "target_node_id": "div@decoder.final_attention#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-96", "source_node_id": "div@decoder.final_attention#0", "target_node_id": "decoder.final_attention.softmax", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-97", "source_node_id": "decoder.final_attention.softmax", "target_node_id": "decoder.final_attention.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-98", "source_node_id": "decoder.final_attention.value", "target_node_id": "decoder.final_attention.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-99", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-100", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-101", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-102", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "add@decoder.second#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-103", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "add@decoder.second#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-104", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-105", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-106", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-107", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "add@decoder.second#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-108", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-109", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-110", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "decoder.second.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-111", "source_node_id": "decoder.first.mlp_norm", "target_node_id": "add@decoder.second#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-112", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-113", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-114", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-115", "source_node_id": "decoder.first.image_norm", "target_node_id": "add@decoder.second#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-116", "source_node_id": "decoder.first.image_norm", "target_node_id": "add@decoder.second#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-117", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-118", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-119", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-120", "source_node_id": "decoder.first.image_norm", "target_node_id": "add@decoder.second#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-121", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-122", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-123", "source_node_id": "decoder.first.image_norm", "target_node_id": "decoder.second.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-124", "source_node_id": "decoder.first.image_norm", "target_node_id": "add@decoder.second#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-125", "source_node_id": "decoder.first", "target_node_id": "decoder.second", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-126", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "add@decoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-127", "source_node_id": "decoder.second.image_norm", "target_node_id": "add@decoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-128", "source_node_id": "decoder.second", "target_node_id": "add@decoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-129", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "add@decoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-130", "source_node_id": "decoder.second.image_norm", "target_node_id": "add@decoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-131", "source_node_id": "decoder.second", "target_node_id": "add@decoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-132", "source_node_id": "add@decoder#1", "target_node_id": "decoder.final_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-133", "source_node_id": "add@decoder#1", "target_node_id": "decoder.final_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-134", "source_node_id": "add@decoder#1", "target_node_id": "decoder.final_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-135", "source_node_id": "add@decoder#1", "target_node_id": "decoder.final_attention", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-136", "source_node_id": "add@decoder#0", "target_node_id": "decoder.final_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-137", "source_node_id": "add@decoder#0", "target_node_id": "decoder.final_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-138", "source_node_id": "add@decoder#0", "target_node_id": "decoder.final_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-139", "source_node_id": "add@decoder#0", "target_node_id": "decoder.final_attention", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-140", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "decoder.final_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-141", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "decoder.final_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-142", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "decoder.final_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-143", "source_node_id": "decoder.second.image_norm", "target_node_id": "decoder.final_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-144", "source_node_id": "decoder.second.image_norm", "target_node_id": "decoder.final_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-145", "source_node_id": "decoder.second.image_norm", "target_node_id": "decoder.final_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-146", "source_node_id": "decoder.second", "target_node_id": "decoder.final_attention", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-147", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "add@decoder#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-148", "source_node_id": "decoder.second.image_norm", "target_node_id": "add@decoder#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-149", "source_node_id": "decoder.second", "target_node_id": "add@decoder#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-150", "source_node_id": "decoder.final_attention.project", "target_node_id": "add@decoder#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-151", "source_node_id": "decoder.final_attention", "target_node_id": "add@decoder#2", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-152", "source_node_id": "add@decoder#2", "target_node_id": "decoder.final_norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-153", "source_node_id": "upscale.first", "target_node_id": "permute@upscale#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-154", "source_node_id": "permute@upscale#0", "target_node_id": "upscale.norm", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-155", "source_node_id": "upscale.norm", "target_node_id": "permute@upscale#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-156", "source_node_id": "permute@upscale#1", "target_node_id": "gelu@upscale#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-157", "source_node_id": "gelu@upscale#0", "target_node_id": "upscale.second", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-158", "source_node_id": "upscale.second", "target_node_id": "gelu@upscale#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-159", "source_node_id": "masks.mask0.first", "target_node_id": "relu@masks.mask0#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-160", "source_node_id": "relu@masks.mask0#0", "target_node_id": "masks.mask0.second", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-161", "source_node_id": "masks.mask0.second", "target_node_id": "relu@masks.mask0#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-162", "source_node_id": "relu@masks.mask0#1", "target_node_id": "masks.mask0.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-163", "source_node_id": "masks.mask1.first", "target_node_id": "relu@masks.mask1#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-164", "source_node_id": "relu@masks.mask1#0", "target_node_id": "masks.mask1.second", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-165", "source_node_id": "masks.mask1.second", "target_node_id": "relu@masks.mask1#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-166", "source_node_id": "relu@masks.mask1#1", "target_node_id": "masks.mask1.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-167", "source_node_id": "masks.mask2.first", "target_node_id": "relu@masks.mask2#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-168", "source_node_id": "relu@masks.mask2#0", "target_node_id": "masks.mask2.second", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-169", "source_node_id": "masks.mask2.second", "target_node_id": "relu@masks.mask2#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-170", "source_node_id": "relu@masks.mask2#1", "target_node_id": "masks.mask2.project", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-171", "source_node_id": "masks.mask0.project", "target_node_id": "stack@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-172", "source_node_id": "masks.mask0", "target_node_id": "stack@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-173", "source_node_id": "masks.mask1.project", "target_node_id": "stack@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-174", "source_node_id": "masks.mask1", "target_node_id": "stack@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-175", "source_node_id": "masks.mask2.project", "target_node_id": "stack@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-176", "source_node_id": "masks.mask2", "target_node_id": "stack@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-177", "source_node_id": "stack@masks#0", "target_node_id": "python@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-178", "source_node_id": "flatten@masks#0", "target_node_id": "python@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-179", "source_node_id": "quality.first", "target_node_id": "relu@quality#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-180", "source_node_id": "relu@quality#0", "target_node_id": "quality.second", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-181", "source_node_id": "quality.second", "target_node_id": "relu@quality#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-182", "source_node_id": "relu@quality#1", "target_node_id": "quality.scores", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-183", "source_node_id": "input", "target_node_id": "cat@prompt_encoder.position#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-184", "source_node_id": "input", "target_node_id": "prompt_encoder.type_embedding", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-185", "source_node_id": "input", "target_node_id": "prompt_encoder", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-186", "source_node_id": "output_tokens", "target_node_id": "cat@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-187", "source_node_id": "input", "target_node_id": "cat@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-188", "source_node_id": "add@prompt_encoder#0", "target_node_id": "cat@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-189", "source_node_id": "prompt_encoder", "target_node_id": "cat@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-190", "source_node_id": "input", "target_node_id": "flatten@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-191", "source_node_id": "flatten@PromptableMaskDecoder#0", "target_node_id": "add@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-192", "source_node_id": "no_mask", "target_node_id": "add@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-193", "source_node_id": "add@prompt_encoder#0", "target_node_id": "python@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-194", "source_node_id": "prompt_encoder", "target_node_id": "python@PromptableMaskDecoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-195", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-196", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-197", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-198", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-199", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-200", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-201", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-202", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-203", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-204", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-205", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-206", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-207", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-208", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-209", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-210", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-211", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-212", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-213", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-214", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-215", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-216", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-217", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-218", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-219", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-220", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-221", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-222", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "add@decoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-223", "source_node_id": "cat@PromptableMaskDecoder#0", "target_node_id": "decoder", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-224", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-225", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-226", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-227", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-228", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-229", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-230", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-231", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-232", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-233", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-234", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-235", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-236", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-237", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-238", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-239", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-240", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-241", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-242", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-243", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-244", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-245", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-246", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-247", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-248", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-249", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-250", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-251", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "add@decoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-252", "source_node_id": "add@PromptableMaskDecoder#0", "target_node_id": "decoder", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-253", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-254", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-255", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-256", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-257", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-258", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-259", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-260", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-261", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-262", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-263", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-264", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.first.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-265", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.first#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-266", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-267", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-268", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.self_attention.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-269", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-270", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-271", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-272", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-273", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.token_to_image.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-274", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#4", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-275", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.query", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-276", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.key", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-277", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder.second.image_to_token.value", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-278", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder.second#5", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-279", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-280", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "add@decoder#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-281", "source_node_id": "python@PromptableMaskDecoder#0", "target_node_id": "decoder", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-282", "source_node_id": "decoder.final_norm", "target_node_id": "upscale.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-283", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "upscale.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-284", "source_node_id": "decoder.second.image_norm", "target_node_id": "upscale.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-285", "source_node_id": "decoder", "target_node_id": "upscale", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-286", "source_node_id": "decoder.final_norm", "target_node_id": "masks.mask0.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-287", "source_node_id": "decoder.final_norm", "target_node_id": "masks.mask1.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-288", "source_node_id": "decoder.final_norm", "target_node_id": "masks.mask2.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-289", "source_node_id": "decoder.final_norm", "target_node_id": "flatten@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-290", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "masks.mask0.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-291", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "masks.mask1.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-292", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "masks.mask2.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-293", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "flatten@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-294", "source_node_id": "decoder.second.image_norm", "target_node_id": "masks.mask0.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-295", "source_node_id": "decoder.second.image_norm", "target_node_id": "masks.mask1.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-296", "source_node_id": "decoder.second.image_norm", "target_node_id": "masks.mask2.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-297", "source_node_id": "decoder.second.image_norm", "target_node_id": "flatten@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-298", "source_node_id": "decoder", "target_node_id": "masks", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-299", "source_node_id": "gelu@upscale#1", "target_node_id": "masks.mask0.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-300", "source_node_id": "gelu@upscale#1", "target_node_id": "masks.mask1.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-301", "source_node_id": "gelu@upscale#1", "target_node_id": "masks.mask2.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-302", "source_node_id": "gelu@upscale#1", "target_node_id": "flatten@masks#0", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-303", "source_node_id": "upscale", "target_node_id": "masks", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-304", "source_node_id": "decoder.final_norm", "target_node_id": "quality.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-305", "source_node_id": "decoder.second.mlp_norm", "target_node_id": "quality.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-306", "source_node_id": "decoder.second.image_norm", "target_node_id": "quality.first", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-307", "source_node_id": "decoder", "target_node_id": "quality", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-308", "source_node_id": "python@masks#0", "target_node_id": "flatten@PromptableMaskDecoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-309", "source_node_id": "masks", "target_node_id": "flatten@PromptableMaskDecoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-310", "source_node_id": "flatten@PromptableMaskDecoder#1", "target_node_id": "cat@PromptableMaskDecoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-311", "source_node_id": "quality.scores", "target_node_id": "cat@PromptableMaskDecoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-312", "source_node_id": "quality", "target_node_id": "cat@PromptableMaskDecoder#1", "source_port": "y", "target_port": "x", "metadata": {} }, { "edge_id": "edge-313", "source_node_id": "cat@PromptableMaskDecoder#1", "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 }, "flatten@PromptableMaskDecoder#0": { "x": 340, "y": 160 }, "prompt_encoder": { "x": 680, "y": 160 }, "python@PromptableMaskDecoder#0": { "x": 1020, "y": 160 }, "output_tokens": { "x": 1360, "y": 160 }, "cat@PromptableMaskDecoder#0": { "x": 1700, "y": 160 }, "no_mask": { "x": 2040, "y": 160 }, "add@PromptableMaskDecoder#0": { "x": 2380, "y": 160 }, "decoder": { "x": 2720, "y": 160 }, "upscale": { "x": 3060, "y": 160 }, "masks": { "x": 3400, "y": 160 }, "flatten@PromptableMaskDecoder#1": { "x": 3740, "y": 160 }, "quality": { "x": 4080, "y": 160 }, "cat@PromptableMaskDecoder#1": { "x": 4420, "y": 160 }, "output": { "x": 4760, "y": 160 }, "prompt_encoder.position": { "x": 24, "y": 214 }, "prompt_encoder.type_embedding": { "x": 24, "y": 364 }, "add@prompt_encoder#0": { "x": 24, "y": 514 }, "cat@prompt_encoder.position#0": { "x": 24, "y": 214 }, "decoder.first": { "x": 24, "y": 214 }, "decoder.second": { "x": 24, "y": 364 }, "add@decoder#0": { "x": 24, "y": 514 }, "add@decoder#1": { "x": 24, "y": 664 }, "decoder.final_attention": { "x": 24, "y": 814 }, "add@decoder#2": { "x": 24, "y": 964 }, "decoder.final_norm": { "x": 24, "y": 1114 }, "decoder.first.self_attention": { "x": 24, "y": 214 }, "add@decoder.first#0": { "x": 24, "y": 364 }, "decoder.first.self_norm": { "x": 24, "y": 514 }, "add@decoder.first#1": { "x": 24, "y": 664 }, "decoder.first.token_to_image": { "x": 24, "y": 814 }, "add@decoder.first#2": { "x": 24, "y": 964 }, "decoder.first.token_norm": { "x": 24, "y": 1114 }, "decoder.first.up": { "x": 24, "y": 1264 }, "decoder.first.relu": { "x": 24, "y": 1414 }, "decoder.first.down": { "x": 24, "y": 1564 }, "add@decoder.first#3": { "x": 24, "y": 1714 }, "decoder.first.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.first#4": { "x": 24, "y": 2014 }, "decoder.first.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.first#5": { "x": 24, "y": 2314 }, "decoder.first.image_norm": { "x": 24, "y": 2464 }, "decoder.first.self_attention.query": { "x": 24, "y": 214 }, "decoder.first.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.first.self_attention#0": { "x": 24, "y": 514 }, "decoder.first.self_attention.value": { "x": 24, "y": 664 }, "decoder.first.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.first.self_attention.project": { "x": 24, "y": 964 }, "decoder.first.token_to_image.query": { "x": 24, "y": 214 }, "decoder.first.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.first.token_to_image#0": { "x": 24, "y": 514 }, "decoder.first.token_to_image.value": { "x": 24, "y": 664 }, "decoder.first.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.first.token_to_image.project": { "x": 24, "y": 964 }, "decoder.first.image_to_token.query": { "x": 24, "y": 214 }, "decoder.first.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.first.image_to_token#0": { "x": 24, "y": 514 }, "decoder.first.image_to_token.value": { "x": 24, "y": 664 }, "decoder.first.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.first.image_to_token.project": { "x": 24, "y": 964 }, "decoder.second.self_attention": { "x": 24, "y": 214 }, "add@decoder.second#0": { "x": 24, "y": 364 }, "decoder.second.self_norm": { "x": 24, "y": 514 }, "add@decoder.second#1": { "x": 24, "y": 664 }, "decoder.second.token_to_image": { "x": 24, "y": 814 }, "add@decoder.second#2": { "x": 24, "y": 964 }, "decoder.second.token_norm": { "x": 24, "y": 1114 }, "decoder.second.up": { "x": 24, "y": 1264 }, "decoder.second.relu": { "x": 24, "y": 1414 }, "decoder.second.down": { "x": 24, "y": 1564 }, "add@decoder.second#3": { "x": 24, "y": 1714 }, "decoder.second.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.second#4": { "x": 24, "y": 2014 }, "decoder.second.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.second#5": { "x": 24, "y": 2314 }, "decoder.second.image_norm": { "x": 24, "y": 2464 }, "decoder.second.self_attention.query": { "x": 24, "y": 214 }, "decoder.second.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.second.self_attention#0": { "x": 24, "y": 514 }, "decoder.second.self_attention.value": { "x": 24, "y": 664 }, "decoder.second.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.second.self_attention.project": { "x": 24, "y": 964 }, "decoder.second.token_to_image.query": { "x": 24, "y": 214 }, "decoder.second.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.second.token_to_image#0": { "x": 24, "y": 514 }, "decoder.second.token_to_image.value": { "x": 24, "y": 664 }, "decoder.second.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.second.token_to_image.project": { "x": 24, "y": 964 }, "decoder.second.image_to_token.query": { "x": 24, "y": 214 }, "decoder.second.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.second.image_to_token#0": { "x": 24, "y": 514 }, "decoder.second.image_to_token.value": { "x": 24, "y": 664 }, "decoder.second.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.second.image_to_token.project": { "x": 24, "y": 964 }, "decoder.final_attention.query": { "x": 24, "y": 214 }, "decoder.final_attention.key": { "x": 24, "y": 364 }, "div@decoder.final_attention#0": { "x": 24, "y": 514 }, "decoder.final_attention.value": { "x": 24, "y": 664 }, "decoder.final_attention.softmax": { "x": 24, "y": 814 }, "decoder.final_attention.project": { "x": 24, "y": 964 }, "upscale.first": { "x": 24, "y": 214 }, "permute@upscale#0": { "x": 24, "y": 364 }, "upscale.norm": { "x": 24, "y": 514 }, "permute@upscale#1": { "x": 24, "y": 664 }, "gelu@upscale#0": { "x": 24, "y": 814 }, "upscale.second": { "x": 24, "y": 964 }, "gelu@upscale#1": { "x": 24, "y": 1114 }, "flatten@masks#0": { "x": 24, "y": 214 }, "masks.mask0": { "x": 24, "y": 364 }, "masks.mask1": { "x": 24, "y": 514 }, "masks.mask2": { "x": 24, "y": 664 }, "stack@masks#0": { "x": 24, "y": 814 }, "python@masks#0": { "x": 24, "y": 964 }, "masks.mask0.first": { "x": 24, "y": 214 }, "relu@masks.mask0#0": { "x": 24, "y": 364 }, "masks.mask0.second": { "x": 24, "y": 514 }, "relu@masks.mask0#1": { "x": 24, "y": 664 }, "masks.mask0.project": { "x": 24, "y": 814 }, "masks.mask1.first": { "x": 24, "y": 214 }, "relu@masks.mask1#0": { "x": 24, "y": 364 }, "masks.mask1.second": { "x": 24, "y": 514 }, "relu@masks.mask1#1": { "x": 24, "y": 664 }, "masks.mask1.project": { "x": 24, "y": 814 }, "masks.mask2.first": { "x": 24, "y": 214 }, "relu@masks.mask2#0": { "x": 24, "y": 364 }, "masks.mask2.second": { "x": 24, "y": 514 }, "relu@masks.mask2#1": { "x": 24, "y": 664 }, "masks.mask2.project": { "x": 24, "y": 814 }, "quality.first": { "x": 24, "y": 214 }, "relu@quality#0": { "x": 24, "y": 364 }, "quality.second": { "x": 24, "y": 514 }, "relu@quality#1": { "x": 24, "y": 664 }, "quality.scores": { "x": 24, "y": 814 } }, "collapsed_modules": [ "prompt_encoder", "decoder", "upscale", "masks", "quality", "prompt_encoder.position", "decoder.first", "decoder.second", "decoder.final_attention", "decoder.first.self_attention", "decoder.first.token_to_image", "decoder.first.image_to_token", "decoder.second.self_attention", "decoder.second.token_to_image", "decoder.second.image_to_token", "masks.mask0", "masks.mask1", "masks.mask2" ], "selection": [], "layout": {} } }, "snapshotId": "snapshot-2b9a0ff6ffe7a61c", "collaboration": { "walkthrough": { "title": "Turn a prompt into candidate masks", "steps": [ { "id": "prompt", "title": "Combine prompt geometry with its meaning", "note": "Random Fourier features encode normalized (x,y) coordinates. A learned type embedding distinguishes foreground/background points and the two box corners. The same image features are reused when a prompt changes.", "view": { "camera": { "x": 308.5, "y": -144, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "flatten@PromptableMaskDecoder#0": { "x": 340, "y": 160 }, "prompt_encoder": { "x": 680, "y": 160 }, "python@PromptableMaskDecoder#0": { "x": 1020, "y": 160 }, "output_tokens": { "x": 1360, "y": 160 }, "cat@PromptableMaskDecoder#0": { "x": 1700, "y": 160 }, "no_mask": { "x": 2040, "y": 160 }, "add@PromptableMaskDecoder#0": { "x": 2380, "y": 160 }, "decoder": { "x": 2720, "y": 160 }, "upscale": { "x": 3060, "y": 160 }, "masks": { "x": 3400, "y": 160 }, "flatten@PromptableMaskDecoder#1": { "x": 3740, "y": 160 }, "quality": { "x": 4080, "y": 160 }, "cat@PromptableMaskDecoder#1": { "x": 4420, "y": 160 }, "output": { "x": 4760, "y": 160 }, "prompt_encoder.position": { "x": 24, "y": 214 }, "prompt_encoder.type_embedding": { "x": 24, "y": 364 }, "add@prompt_encoder#0": { "x": 24, "y": 514 }, "cat@prompt_encoder.position#0": { "x": 24, "y": 214 }, "decoder.first": { "x": 24, "y": 214 }, "decoder.second": { "x": 24, "y": 364 }, "add@decoder#0": { "x": 24, "y": 514 }, "add@decoder#1": { "x": 24, "y": 664 }, "decoder.final_attention": { "x": 24, "y": 814 }, "add@decoder#2": { "x": 24, "y": 964 }, "decoder.final_norm": { "x": 24, "y": 1114 }, "decoder.first.self_attention": { "x": 24, "y": 214 }, "add@decoder.first#0": { "x": 24, "y": 364 }, "decoder.first.self_norm": { "x": 24, "y": 514 }, "add@decoder.first#1": { "x": 24, "y": 664 }, "decoder.first.token_to_image": { "x": 24, "y": 814 }, "add@decoder.first#2": { "x": 24, "y": 964 }, "decoder.first.token_norm": { "x": 24, "y": 1114 }, "decoder.first.up": { "x": 24, "y": 1264 }, "decoder.first.relu": { "x": 24, "y": 1414 }, "decoder.first.down": { "x": 24, "y": 1564 }, "add@decoder.first#3": { "x": 24, "y": 1714 }, "decoder.first.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.first#4": { "x": 24, "y": 2014 }, "decoder.first.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.first#5": { "x": 24, "y": 2314 }, "decoder.first.image_norm": { "x": 24, "y": 2464 }, "decoder.first.self_attention.query": { "x": 24, "y": 214 }, "decoder.first.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.first.self_attention#0": { "x": 24, "y": 514 }, "decoder.first.self_attention.value": { "x": 24, "y": 664 }, "decoder.first.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.first.self_attention.project": { "x": 24, "y": 964 }, "decoder.first.token_to_image.query": { "x": 24, "y": 214 }, "decoder.first.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.first.token_to_image#0": { "x": 24, "y": 514 }, "decoder.first.token_to_image.value": { "x": 24, "y": 664 }, "decoder.first.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.first.token_to_image.project": { "x": 24, "y": 964 }, "decoder.first.image_to_token.query": { "x": 24, "y": 214 }, "decoder.first.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.first.image_to_token#0": { "x": 24, "y": 514 }, "decoder.first.image_to_token.value": { "x": 24, "y": 664 }, "decoder.first.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.first.image_to_token.project": { "x": 24, "y": 964 }, "decoder.second.self_attention": { "x": 24, "y": 214 }, "add@decoder.second#0": { "x": 24, "y": 364 }, "decoder.second.self_norm": { "x": 24, "y": 514 }, "add@decoder.second#1": { "x": 24, "y": 664 }, "decoder.second.token_to_image": { "x": 24, "y": 814 }, "add@decoder.second#2": { "x": 24, "y": 964 }, "decoder.second.token_norm": { "x": 24, "y": 1114 }, "decoder.second.up": { "x": 24, "y": 1264 }, "decoder.second.relu": { "x": 24, "y": 1414 }, "decoder.second.down": { "x": 24, "y": 1564 }, "add@decoder.second#3": { "x": 24, "y": 1714 }, "decoder.second.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.second#4": { "x": 24, "y": 2014 }, "decoder.second.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.second#5": { "x": 24, "y": 2314 }, "decoder.second.image_norm": { "x": 24, "y": 2464 }, "decoder.second.self_attention.query": { "x": 24, "y": 214 }, "decoder.second.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.second.self_attention#0": { "x": 24, "y": 514 }, "decoder.second.self_attention.value": { "x": 24, "y": 664 }, "decoder.second.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.second.self_attention.project": { "x": 24, "y": 964 }, "decoder.second.token_to_image.query": { "x": 24, "y": 214 }, "decoder.second.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.second.token_to_image#0": { "x": 24, "y": 514 }, "decoder.second.token_to_image.value": { "x": 24, "y": 664 }, "decoder.second.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.second.token_to_image.project": { "x": 24, "y": 964 }, "decoder.second.image_to_token.query": { "x": 24, "y": 214 }, "decoder.second.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.second.image_to_token#0": { "x": 24, "y": 514 }, "decoder.second.image_to_token.value": { "x": 24, "y": 664 }, "decoder.second.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.second.image_to_token.project": { "x": 24, "y": 964 }, "decoder.final_attention.query": { "x": 24, "y": 214 }, "decoder.final_attention.key": { "x": 24, "y": 364 }, "div@decoder.final_attention#0": { "x": 24, "y": 514 }, "decoder.final_attention.value": { "x": 24, "y": 664 }, "decoder.final_attention.softmax": { "x": 24, "y": 814 }, "decoder.final_attention.project": { "x": 24, "y": 964 }, "upscale.first": { "x": 24, "y": 214 }, "permute@upscale#0": { "x": 24, "y": 364 }, "upscale.norm": { "x": 24, "y": 514 }, "permute@upscale#1": { "x": 24, "y": 664 }, "gelu@upscale#0": { "x": 24, "y": 814 }, "upscale.second": { "x": 24, "y": 964 }, "gelu@upscale#1": { "x": 24, "y": 1114 }, "flatten@masks#0": { "x": 24, "y": 214 }, "masks.mask0": { "x": 24, "y": 364 }, "masks.mask1": { "x": 24, "y": 514 }, "masks.mask2": { "x": 24, "y": 664 }, "stack@masks#0": { "x": 24, "y": 814 }, "python@masks#0": { "x": 24, "y": 964 }, "masks.mask0.first": { "x": 24, "y": 214 }, "relu@masks.mask0#0": { "x": 24, "y": 364 }, "masks.mask0.second": { "x": 24, "y": 514 }, "relu@masks.mask0#1": { "x": 24, "y": 664 }, "masks.mask0.project": { "x": 24, "y": 814 }, "masks.mask1.first": { "x": 24, "y": 214 }, "relu@masks.mask1#0": { "x": 24, "y": 364 }, "masks.mask1.second": { "x": 24, "y": 514 }, "relu@masks.mask1#1": { "x": 24, "y": 664 }, "masks.mask1.project": { "x": 24, "y": 814 }, "masks.mask2.first": { "x": 24, "y": 214 }, "relu@masks.mask2#0": { "x": 24, "y": 364 }, "masks.mask2.second": { "x": 24, "y": 514 }, "relu@masks.mask2#1": { "x": 24, "y": 664 }, "masks.mask2.project": { "x": 24, "y": 814 }, "quality.first": { "x": 24, "y": 214 }, "relu@quality#0": { "x": 24, "y": 364 }, "quality.second": { "x": 24, "y": 514 }, "relu@quality#1": { "x": 24, "y": 664 }, "quality.scores": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "prompt_encoder" ], "selectedNodeIds": [ "prompt_encoder.type_embedding" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "prompt_encoder.type_embedding" }, "snapshotId": "snapshot-2b9a0ff6ffe7a61c", "exportedAt": "2026-09-17T07:00:38.449307+00:00", "modelName": "PromptableMaskDecoder", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"prompt_encoder.type_embedding\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"prompt_encoder.type_embedding\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@prompt_encoder#0\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[],\"node\":{\"id\":\"prompt_encoder.type_embedding\",\"kind\":\"layer\",\"op\":\"Embedding\",\"params\":{\"embedding_dim\":8,\"num_embeddings\":4},\"parent\":\"prompt_encoder\"}}" } ] }, { "id": "tokens", "title": "Let prompt and output tokens read image features", "note": "Three mask tokens, one quality token and the sparse prompts pass through self-attention, token-to-image attention and an MLP. Prompt and image positions are added to attention queries and keys.", "view": { "camera": { "x": 291, "y": -604, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "flatten@PromptableMaskDecoder#0": { "x": 340, "y": 160 }, "prompt_encoder": { "x": 680, "y": 160 }, "python@PromptableMaskDecoder#0": { "x": 1020, "y": 160 }, "output_tokens": { "x": 1360, "y": 160 }, "cat@PromptableMaskDecoder#0": { "x": 1700, "y": 160 }, "no_mask": { "x": 2040, "y": 160 }, "add@PromptableMaskDecoder#0": { "x": 2380, "y": 160 }, "decoder": { "x": 2720, "y": 160 }, "upscale": { "x": 3060, "y": 160 }, "masks": { "x": 3400, "y": 160 }, "flatten@PromptableMaskDecoder#1": { "x": 3740, "y": 160 }, "quality": { "x": 4080, "y": 160 }, "cat@PromptableMaskDecoder#1": { "x": 4420, "y": 160 }, "output": { "x": 4760, "y": 160 }, "prompt_encoder.position": { "x": 24, "y": 214 }, "prompt_encoder.type_embedding": { "x": 24, "y": 364 }, "add@prompt_encoder#0": { "x": 24, "y": 514 }, "cat@prompt_encoder.position#0": { "x": 24, "y": 214 }, "decoder.first": { "x": 24, "y": 214 }, "decoder.second": { "x": 24, "y": 364 }, "add@decoder#0": { "x": 24, "y": 514 }, "add@decoder#1": { "x": 24, "y": 664 }, "decoder.final_attention": { "x": 24, "y": 814 }, "add@decoder#2": { "x": 24, "y": 964 }, "decoder.final_norm": { "x": 24, "y": 1114 }, "decoder.first.self_attention": { "x": 24, "y": 214 }, "add@decoder.first#0": { "x": 24, "y": 364 }, "decoder.first.self_norm": { "x": 24, "y": 514 }, "add@decoder.first#1": { "x": 24, "y": 664 }, "decoder.first.token_to_image": { "x": 24, "y": 814 }, "add@decoder.first#2": { "x": 24, "y": 964 }, "decoder.first.token_norm": { "x": 24, "y": 1114 }, "decoder.first.up": { "x": 24, "y": 1264 }, "decoder.first.relu": { "x": 24, "y": 1414 }, "decoder.first.down": { "x": 24, "y": 1564 }, "add@decoder.first#3": { "x": 24, "y": 1714 }, "decoder.first.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.first#4": { "x": 24, "y": 2014 }, "decoder.first.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.first#5": { "x": 24, "y": 2314 }, "decoder.first.image_norm": { "x": 24, "y": 2464 }, "decoder.first.self_attention.query": { "x": 24, "y": 214 }, "decoder.first.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.first.self_attention#0": { "x": 24, "y": 514 }, "decoder.first.self_attention.value": { "x": 24, "y": 664 }, "decoder.first.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.first.self_attention.project": { "x": 24, "y": 964 }, "decoder.first.token_to_image.query": { "x": 24, "y": 214 }, "decoder.first.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.first.token_to_image#0": { "x": 24, "y": 514 }, "decoder.first.token_to_image.value": { "x": 24, "y": 664 }, "decoder.first.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.first.token_to_image.project": { "x": 24, "y": 964 }, "decoder.first.image_to_token.query": { "x": 24, "y": 214 }, "decoder.first.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.first.image_to_token#0": { "x": 24, "y": 514 }, "decoder.first.image_to_token.value": { "x": 24, "y": 664 }, "decoder.first.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.first.image_to_token.project": { "x": 24, "y": 964 }, "decoder.second.self_attention": { "x": 24, "y": 214 }, "add@decoder.second#0": { "x": 24, "y": 364 }, "decoder.second.self_norm": { "x": 24, "y": 514 }, "add@decoder.second#1": { "x": 24, "y": 664 }, "decoder.second.token_to_image": { "x": 24, "y": 814 }, "add@decoder.second#2": { "x": 24, "y": 964 }, "decoder.second.token_norm": { "x": 24, "y": 1114 }, "decoder.second.up": { "x": 24, "y": 1264 }, "decoder.second.relu": { "x": 24, "y": 1414 }, "decoder.second.down": { "x": 24, "y": 1564 }, "add@decoder.second#3": { "x": 24, "y": 1714 }, "decoder.second.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.second#4": { "x": 24, "y": 2014 }, "decoder.second.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.second#5": { "x": 24, "y": 2314 }, "decoder.second.image_norm": { "x": 24, "y": 2464 }, "decoder.second.self_attention.query": { "x": 24, "y": 214 }, "decoder.second.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.second.self_attention#0": { "x": 24, "y": 514 }, "decoder.second.self_attention.value": { "x": 24, "y": 664 }, "decoder.second.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.second.self_attention.project": { "x": 24, "y": 964 }, "decoder.second.token_to_image.query": { "x": 24, "y": 214 }, "decoder.second.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.second.token_to_image#0": { "x": 24, "y": 514 }, "decoder.second.token_to_image.value": { "x": 24, "y": 664 }, "decoder.second.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.second.token_to_image.project": { "x": 24, "y": 964 }, "decoder.second.image_to_token.query": { "x": 24, "y": 214 }, "decoder.second.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.second.image_to_token#0": { "x": 24, "y": 514 }, "decoder.second.image_to_token.value": { "x": 24, "y": 664 }, "decoder.second.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.second.image_to_token.project": { "x": 24, "y": 964 }, "decoder.final_attention.query": { "x": 24, "y": 214 }, "decoder.final_attention.key": { "x": 24, "y": 364 }, "div@decoder.final_attention#0": { "x": 24, "y": 514 }, "decoder.final_attention.value": { "x": 24, "y": 664 }, "decoder.final_attention.softmax": { "x": 24, "y": 814 }, "decoder.final_attention.project": { "x": 24, "y": 964 }, "upscale.first": { "x": 24, "y": 214 }, "permute@upscale#0": { "x": 24, "y": 364 }, "upscale.norm": { "x": 24, "y": 514 }, "permute@upscale#1": { "x": 24, "y": 664 }, "gelu@upscale#0": { "x": 24, "y": 814 }, "upscale.second": { "x": 24, "y": 964 }, "gelu@upscale#1": { "x": 24, "y": 1114 }, "flatten@masks#0": { "x": 24, "y": 214 }, "masks.mask0": { "x": 24, "y": 364 }, "masks.mask1": { "x": 24, "y": 514 }, "masks.mask2": { "x": 24, "y": 664 }, "stack@masks#0": { "x": 24, "y": 814 }, "python@masks#0": { "x": 24, "y": 964 }, "masks.mask0.first": { "x": 24, "y": 214 }, "relu@masks.mask0#0": { "x": 24, "y": 364 }, "masks.mask0.second": { "x": 24, "y": 514 }, "relu@masks.mask0#1": { "x": 24, "y": 664 }, "masks.mask0.project": { "x": 24, "y": 814 }, "masks.mask1.first": { "x": 24, "y": 214 }, "relu@masks.mask1#0": { "x": 24, "y": 364 }, "masks.mask1.second": { "x": 24, "y": 514 }, "relu@masks.mask1#1": { "x": 24, "y": 664 }, "masks.mask1.project": { "x": 24, "y": 814 }, "masks.mask2.first": { "x": 24, "y": 214 }, "relu@masks.mask2#0": { "x": 24, "y": 364 }, "masks.mask2.second": { "x": 24, "y": 514 }, "relu@masks.mask2#1": { "x": 24, "y": 664 }, "masks.mask2.project": { "x": 24, "y": 814 }, "quality.first": { "x": 24, "y": 214 }, "relu@quality#0": { "x": 24, "y": 364 }, "quality.second": { "x": 24, "y": 514 }, "relu@quality#1": { "x": 24, "y": 664 }, "quality.scores": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "decoder", "decoder.first" ], "selectedNodeIds": [ "decoder.first.token_to_image" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "decoder.first.token_to_image" }, "snapshotId": "snapshot-2b9a0ff6ffe7a61c", "exportedAt": "2026-09-17T07:00:38.449307+00:00", "modelName": "PromptableMaskDecoder", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.value\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.value\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"cat@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"cat@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"cat@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.value\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.token_to_image\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@decoder.first#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.token_to_image.key\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"div@decoder.first.token_to_image#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.token_to_image.project\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@decoder.first#2\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.token_to_image.query\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"div@decoder.first.token_to_image#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.token_to_image.softmax\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.project\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.token_to_image.value\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.project\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"div@decoder.first.token_to_image#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.softmax\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.token_to_image.value\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[\"{\\\"id\\\":\\\"decoder.first.token_to_image.key\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":8,\\\"out_features\\\":4},\\\"parent\\\":\\\"decoder.first.token_to_image\\\"}\",\"{\\\"id\\\":\\\"decoder.first.token_to_image.project\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":4,\\\"out_features\\\":8},\\\"parent\\\":\\\"decoder.first.token_to_image\\\"}\",\"{\\\"id\\\":\\\"decoder.first.token_to_image.query\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":8,\\\"out_features\\\":4},\\\"parent\\\":\\\"decoder.first.token_to_image\\\"}\",\"{\\\"id\\\":\\\"decoder.first.token_to_image.softmax\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Softmax\\\",\\\"params\\\":{},\\\"parent\\\":\\\"decoder.first.token_to_image\\\"}\",\"{\\\"id\\\":\\\"decoder.first.token_to_image.value\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":8,\\\"out_features\\\":4},\\\"parent\\\":\\\"decoder.first.token_to_image\\\"}\",\"{\\\"id\\\":\\\"div@decoder.first.token_to_image#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.div\\\",\\\"params\\\":{},\\\"parent\\\":\\\"decoder.first.token_to_image\\\"}\"],\"node\":{\"id\":\"decoder.first.token_to_image\",\"kind\":\"module\",\"op\":\"Module\",\"params\":{},\"parent\":\"decoder.first\"}}" } ] }, { "id": "image", "title": "Update image features with prompt information", "note": "Attention also runs in the reverse direction: image locations query the updated tokens. A second two-way block repeats this exchange, followed by a final token-to-image read. Both representation streams can now depend on the prompt.", "view": { "camera": { "x": 291, "y": -1954, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "flatten@PromptableMaskDecoder#0": { "x": 340, "y": 160 }, "prompt_encoder": { "x": 680, "y": 160 }, "python@PromptableMaskDecoder#0": { "x": 1020, "y": 160 }, "output_tokens": { "x": 1360, "y": 160 }, "cat@PromptableMaskDecoder#0": { "x": 1700, "y": 160 }, "no_mask": { "x": 2040, "y": 160 }, "add@PromptableMaskDecoder#0": { "x": 2380, "y": 160 }, "decoder": { "x": 2720, "y": 160 }, "upscale": { "x": 3060, "y": 160 }, "masks": { "x": 3400, "y": 160 }, "flatten@PromptableMaskDecoder#1": { "x": 3740, "y": 160 }, "quality": { "x": 4080, "y": 160 }, "cat@PromptableMaskDecoder#1": { "x": 4420, "y": 160 }, "output": { "x": 4760, "y": 160 }, "prompt_encoder.position": { "x": 24, "y": 214 }, "prompt_encoder.type_embedding": { "x": 24, "y": 364 }, "add@prompt_encoder#0": { "x": 24, "y": 514 }, "cat@prompt_encoder.position#0": { "x": 24, "y": 214 }, "decoder.first": { "x": 24, "y": 214 }, "decoder.second": { "x": 24, "y": 364 }, "add@decoder#0": { "x": 24, "y": 514 }, "add@decoder#1": { "x": 24, "y": 664 }, "decoder.final_attention": { "x": 24, "y": 814 }, "add@decoder#2": { "x": 24, "y": 964 }, "decoder.final_norm": { "x": 24, "y": 1114 }, "decoder.first.self_attention": { "x": 24, "y": 214 }, "add@decoder.first#0": { "x": 24, "y": 364 }, "decoder.first.self_norm": { "x": 24, "y": 514 }, "add@decoder.first#1": { "x": 24, "y": 664 }, "decoder.first.token_to_image": { "x": 24, "y": 814 }, "add@decoder.first#2": { "x": 24, "y": 964 }, "decoder.first.token_norm": { "x": 24, "y": 1114 }, "decoder.first.up": { "x": 24, "y": 1264 }, "decoder.first.relu": { "x": 24, "y": 1414 }, "decoder.first.down": { "x": 24, "y": 1564 }, "add@decoder.first#3": { "x": 24, "y": 1714 }, "decoder.first.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.first#4": { "x": 24, "y": 2014 }, "decoder.first.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.first#5": { "x": 24, "y": 2314 }, "decoder.first.image_norm": { "x": 24, "y": 2464 }, "decoder.first.self_attention.query": { "x": 24, "y": 214 }, "decoder.first.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.first.self_attention#0": { "x": 24, "y": 514 }, "decoder.first.self_attention.value": { "x": 24, "y": 664 }, "decoder.first.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.first.self_attention.project": { "x": 24, "y": 964 }, "decoder.first.token_to_image.query": { "x": 24, "y": 214 }, "decoder.first.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.first.token_to_image#0": { "x": 24, "y": 514 }, "decoder.first.token_to_image.value": { "x": 24, "y": 664 }, "decoder.first.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.first.token_to_image.project": { "x": 24, "y": 964 }, "decoder.first.image_to_token.query": { "x": 24, "y": 214 }, "decoder.first.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.first.image_to_token#0": { "x": 24, "y": 514 }, "decoder.first.image_to_token.value": { "x": 24, "y": 664 }, "decoder.first.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.first.image_to_token.project": { "x": 24, "y": 964 }, "decoder.second.self_attention": { "x": 24, "y": 214 }, "add@decoder.second#0": { "x": 24, "y": 364 }, "decoder.second.self_norm": { "x": 24, "y": 514 }, "add@decoder.second#1": { "x": 24, "y": 664 }, "decoder.second.token_to_image": { "x": 24, "y": 814 }, "add@decoder.second#2": { "x": 24, "y": 964 }, "decoder.second.token_norm": { "x": 24, "y": 1114 }, "decoder.second.up": { "x": 24, "y": 1264 }, "decoder.second.relu": { "x": 24, "y": 1414 }, "decoder.second.down": { "x": 24, "y": 1564 }, "add@decoder.second#3": { "x": 24, "y": 1714 }, "decoder.second.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.second#4": { "x": 24, "y": 2014 }, "decoder.second.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.second#5": { "x": 24, "y": 2314 }, "decoder.second.image_norm": { "x": 24, "y": 2464 }, "decoder.second.self_attention.query": { "x": 24, "y": 214 }, "decoder.second.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.second.self_attention#0": { "x": 24, "y": 514 }, "decoder.second.self_attention.value": { "x": 24, "y": 664 }, "decoder.second.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.second.self_attention.project": { "x": 24, "y": 964 }, "decoder.second.token_to_image.query": { "x": 24, "y": 214 }, "decoder.second.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.second.token_to_image#0": { "x": 24, "y": 514 }, "decoder.second.token_to_image.value": { "x": 24, "y": 664 }, "decoder.second.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.second.token_to_image.project": { "x": 24, "y": 964 }, "decoder.second.image_to_token.query": { "x": 24, "y": 214 }, "decoder.second.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.second.image_to_token#0": { "x": 24, "y": 514 }, "decoder.second.image_to_token.value": { "x": 24, "y": 664 }, "decoder.second.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.second.image_to_token.project": { "x": 24, "y": 964 }, "decoder.final_attention.query": { "x": 24, "y": 214 }, "decoder.final_attention.key": { "x": 24, "y": 364 }, "div@decoder.final_attention#0": { "x": 24, "y": 514 }, "decoder.final_attention.value": { "x": 24, "y": 664 }, "decoder.final_attention.softmax": { "x": 24, "y": 814 }, "decoder.final_attention.project": { "x": 24, "y": 964 }, "upscale.first": { "x": 24, "y": 214 }, "permute@upscale#0": { "x": 24, "y": 364 }, "upscale.norm": { "x": 24, "y": 514 }, "permute@upscale#1": { "x": 24, "y": 664 }, "gelu@upscale#0": { "x": 24, "y": 814 }, "upscale.second": { "x": 24, "y": 964 }, "gelu@upscale#1": { "x": 24, "y": 1114 }, "flatten@masks#0": { "x": 24, "y": 214 }, "masks.mask0": { "x": 24, "y": 364 }, "masks.mask1": { "x": 24, "y": 514 }, "masks.mask2": { "x": 24, "y": 664 }, "stack@masks#0": { "x": 24, "y": 814 }, "python@masks#0": { "x": 24, "y": 964 }, "masks.mask0.first": { "x": 24, "y": 214 }, "relu@masks.mask0#0": { "x": 24, "y": 364 }, "masks.mask0.second": { "x": 24, "y": 514 }, "relu@masks.mask0#1": { "x": 24, "y": 664 }, "masks.mask0.project": { "x": 24, "y": 814 }, "masks.mask1.first": { "x": 24, "y": 214 }, "relu@masks.mask1#0": { "x": 24, "y": 364 }, "masks.mask1.second": { "x": 24, "y": 514 }, "relu@masks.mask1#1": { "x": 24, "y": 664 }, "masks.mask1.project": { "x": 24, "y": 814 }, "masks.mask2.first": { "x": 24, "y": 214 }, "relu@masks.mask2#0": { "x": 24, "y": 364 }, "masks.mask2.second": { "x": 24, "y": 514 }, "relu@masks.mask2#1": { "x": 24, "y": 664 }, "masks.mask2.project": { "x": 24, "y": 814 }, "quality.first": { "x": 24, "y": 214 }, "relu@quality#0": { "x": 24, "y": 364 }, "quality.second": { "x": 24, "y": 514 }, "relu@quality#1": { "x": 24, "y": 664 }, "quality.scores": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "decoder", "decoder.first" ], "selectedNodeIds": [ "decoder.first.image_to_token" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "decoder.first.image_to_token" }, "snapshotId": "snapshot-2b9a0ff6ffe7a61c", "exportedAt": "2026-09-17T07:00:38.449307+00:00", "modelName": "PromptableMaskDecoder", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.value\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#4\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#4\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#4\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"add@decoder.first#4\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.value\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"cat@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"cat@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"cat@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.value\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.image_to_token\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@decoder.first#5\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.image_to_token.key\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"div@decoder.first.image_to_token#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.image_to_token.project\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@decoder.first#5\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.image_to_token.query\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"div@decoder.first.image_to_token#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.image_to_token.softmax\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.project\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.image_to_token.value\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.project\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.mlp_norm\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.mlp_norm\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.mlp_norm\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"decoder.first.mlp_norm\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.value\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"div@decoder.first.image_to_token#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.softmax\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.key\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.query\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"python@PromptableMaskDecoder#0\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"decoder.first.image_to_token.value\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[\"{\\\"id\\\":\\\"decoder.first.image_to_token.key\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":8,\\\"out_features\\\":4},\\\"parent\\\":\\\"decoder.first.image_to_token\\\"}\",\"{\\\"id\\\":\\\"decoder.first.image_to_token.project\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":4,\\\"out_features\\\":8},\\\"parent\\\":\\\"decoder.first.image_to_token\\\"}\",\"{\\\"id\\\":\\\"decoder.first.image_to_token.query\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":8,\\\"out_features\\\":4},\\\"parent\\\":\\\"decoder.first.image_to_token\\\"}\",\"{\\\"id\\\":\\\"decoder.first.image_to_token.softmax\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Softmax\\\",\\\"params\\\":{},\\\"parent\\\":\\\"decoder.first.image_to_token\\\"}\",\"{\\\"id\\\":\\\"decoder.first.image_to_token.value\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"Linear\\\",\\\"params\\\":{\\\"in_features\\\":8,\\\"out_features\\\":4},\\\"parent\\\":\\\"decoder.first.image_to_token\\\"}\",\"{\\\"id\\\":\\\"div@decoder.first.image_to_token#0\\\",\\\"kind\\\":\\\"layer\\\",\\\"op\\\":\\\"torch.div\\\",\\\"params\\\":{},\\\"parent\\\":\\\"decoder.first.image_to_token\\\"}\"],\"node\":{\"id\":\"decoder.first.image_to_token\",\"kind\":\"module\",\"op\":\"Module\",\"params\":{},\"parent\":\"decoder.first\"}}" } ] }, { "id": "masks", "title": "Use mask tokens as dynamic pixel classifiers", "note": "Each mask token passes through its own MLP to produce two feature weights. A dot product with every upscaled image location creates one candidate mask logit. A separate quality head scores the three candidates; none of these random outputs is a trained segmentation result.", "view": { "camera": { "x": 308.5, "y": -594, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "flatten@PromptableMaskDecoder#0": { "x": 340, "y": 160 }, "prompt_encoder": { "x": 680, "y": 160 }, "python@PromptableMaskDecoder#0": { "x": 1020, "y": 160 }, "output_tokens": { "x": 1360, "y": 160 }, "cat@PromptableMaskDecoder#0": { "x": 1700, "y": 160 }, "no_mask": { "x": 2040, "y": 160 }, "add@PromptableMaskDecoder#0": { "x": 2380, "y": 160 }, "decoder": { "x": 2720, "y": 160 }, "upscale": { "x": 3060, "y": 160 }, "masks": { "x": 3400, "y": 160 }, "flatten@PromptableMaskDecoder#1": { "x": 3740, "y": 160 }, "quality": { "x": 4080, "y": 160 }, "cat@PromptableMaskDecoder#1": { "x": 4420, "y": 160 }, "output": { "x": 4760, "y": 160 }, "prompt_encoder.position": { "x": 24, "y": 214 }, "prompt_encoder.type_embedding": { "x": 24, "y": 364 }, "add@prompt_encoder#0": { "x": 24, "y": 514 }, "cat@prompt_encoder.position#0": { "x": 24, "y": 214 }, "decoder.first": { "x": 24, "y": 214 }, "decoder.second": { "x": 24, "y": 364 }, "add@decoder#0": { "x": 24, "y": 514 }, "add@decoder#1": { "x": 24, "y": 664 }, "decoder.final_attention": { "x": 24, "y": 814 }, "add@decoder#2": { "x": 24, "y": 964 }, "decoder.final_norm": { "x": 24, "y": 1114 }, "decoder.first.self_attention": { "x": 24, "y": 214 }, "add@decoder.first#0": { "x": 24, "y": 364 }, "decoder.first.self_norm": { "x": 24, "y": 514 }, "add@decoder.first#1": { "x": 24, "y": 664 }, "decoder.first.token_to_image": { "x": 24, "y": 814 }, "add@decoder.first#2": { "x": 24, "y": 964 }, "decoder.first.token_norm": { "x": 24, "y": 1114 }, "decoder.first.up": { "x": 24, "y": 1264 }, "decoder.first.relu": { "x": 24, "y": 1414 }, "decoder.first.down": { "x": 24, "y": 1564 }, "add@decoder.first#3": { "x": 24, "y": 1714 }, "decoder.first.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.first#4": { "x": 24, "y": 2014 }, "decoder.first.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.first#5": { "x": 24, "y": 2314 }, "decoder.first.image_norm": { "x": 24, "y": 2464 }, "decoder.first.self_attention.query": { "x": 24, "y": 214 }, "decoder.first.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.first.self_attention#0": { "x": 24, "y": 514 }, "decoder.first.self_attention.value": { "x": 24, "y": 664 }, "decoder.first.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.first.self_attention.project": { "x": 24, "y": 964 }, "decoder.first.token_to_image.query": { "x": 24, "y": 214 }, "decoder.first.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.first.token_to_image#0": { "x": 24, "y": 514 }, "decoder.first.token_to_image.value": { "x": 24, "y": 664 }, "decoder.first.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.first.token_to_image.project": { "x": 24, "y": 964 }, "decoder.first.image_to_token.query": { "x": 24, "y": 214 }, "decoder.first.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.first.image_to_token#0": { "x": 24, "y": 514 }, "decoder.first.image_to_token.value": { "x": 24, "y": 664 }, "decoder.first.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.first.image_to_token.project": { "x": 24, "y": 964 }, "decoder.second.self_attention": { "x": 24, "y": 214 }, "add@decoder.second#0": { "x": 24, "y": 364 }, "decoder.second.self_norm": { "x": 24, "y": 514 }, "add@decoder.second#1": { "x": 24, "y": 664 }, "decoder.second.token_to_image": { "x": 24, "y": 814 }, "add@decoder.second#2": { "x": 24, "y": 964 }, "decoder.second.token_norm": { "x": 24, "y": 1114 }, "decoder.second.up": { "x": 24, "y": 1264 }, "decoder.second.relu": { "x": 24, "y": 1414 }, "decoder.second.down": { "x": 24, "y": 1564 }, "add@decoder.second#3": { "x": 24, "y": 1714 }, "decoder.second.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.second#4": { "x": 24, "y": 2014 }, "decoder.second.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.second#5": { "x": 24, "y": 2314 }, "decoder.second.image_norm": { "x": 24, "y": 2464 }, "decoder.second.self_attention.query": { "x": 24, "y": 214 }, "decoder.second.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.second.self_attention#0": { "x": 24, "y": 514 }, "decoder.second.self_attention.value": { "x": 24, "y": 664 }, "decoder.second.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.second.self_attention.project": { "x": 24, "y": 964 }, "decoder.second.token_to_image.query": { "x": 24, "y": 214 }, "decoder.second.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.second.token_to_image#0": { "x": 24, "y": 514 }, "decoder.second.token_to_image.value": { "x": 24, "y": 664 }, "decoder.second.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.second.token_to_image.project": { "x": 24, "y": 964 }, "decoder.second.image_to_token.query": { "x": 24, "y": 214 }, "decoder.second.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.second.image_to_token#0": { "x": 24, "y": 514 }, "decoder.second.image_to_token.value": { "x": 24, "y": 664 }, "decoder.second.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.second.image_to_token.project": { "x": 24, "y": 964 }, "decoder.final_attention.query": { "x": 24, "y": 214 }, "decoder.final_attention.key": { "x": 24, "y": 364 }, "div@decoder.final_attention#0": { "x": 24, "y": 514 }, "decoder.final_attention.value": { "x": 24, "y": 664 }, "decoder.final_attention.softmax": { "x": 24, "y": 814 }, "decoder.final_attention.project": { "x": 24, "y": 964 }, "upscale.first": { "x": 24, "y": 214 }, "permute@upscale#0": { "x": 24, "y": 364 }, "upscale.norm": { "x": 24, "y": 514 }, "permute@upscale#1": { "x": 24, "y": 664 }, "gelu@upscale#0": { "x": 24, "y": 814 }, "upscale.second": { "x": 24, "y": 964 }, "gelu@upscale#1": { "x": 24, "y": 1114 }, "flatten@masks#0": { "x": 24, "y": 214 }, "masks.mask0": { "x": 24, "y": 364 }, "masks.mask1": { "x": 24, "y": 514 }, "masks.mask2": { "x": 24, "y": 664 }, "stack@masks#0": { "x": 24, "y": 814 }, "python@masks#0": { "x": 24, "y": 964 }, "masks.mask0.first": { "x": 24, "y": 214 }, "relu@masks.mask0#0": { "x": 24, "y": 364 }, "masks.mask0.second": { "x": 24, "y": 514 }, "relu@masks.mask0#1": { "x": 24, "y": 664 }, "masks.mask0.project": { "x": 24, "y": 814 }, "masks.mask1.first": { "x": 24, "y": 214 }, "relu@masks.mask1#0": { "x": 24, "y": 364 }, "masks.mask1.second": { "x": 24, "y": 514 }, "relu@masks.mask1#1": { "x": 24, "y": 664 }, "masks.mask1.project": { "x": 24, "y": 814 }, "masks.mask2.first": { "x": 24, "y": 214 }, "relu@masks.mask2#0": { "x": 24, "y": 364 }, "masks.mask2.second": { "x": 24, "y": 514 }, "relu@masks.mask2#1": { "x": 24, "y": 664 }, "masks.mask2.project": { "x": 24, "y": 814 }, "quality.first": { "x": 24, "y": 214 }, "relu@quality#0": { "x": 24, "y": 364 }, "quality.second": { "x": 24, "y": 514 }, "relu@quality#1": { "x": 24, "y": 664 }, "quality.scores": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "masks", "masks.mask0" ], "selectedNodeIds": [ "masks.mask0.project" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "masks.mask0.project" }, "snapshotId": "snapshot-2b9a0ff6ffe7a61c", "exportedAt": "2026-09-17T07:00:38.449307+00:00", "modelName": "PromptableMaskDecoder", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"masks.mask0.project\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"stack@masks#0\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"relu@masks.mask0#1\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"masks.mask0.project\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[],\"node\":{\"id\":\"masks.mask0.project\",\"kind\":\"layer\",\"op\":\"Linear\",\"params\":{\"in_features\":8,\"out_features\":2},\"parent\":\"masks.mask0\"}}" } ] } ] }, "threads": [], "savedViews": [ { "id": "prompt", "title": "Combine prompt geometry with its meaning", "note": "Random Fourier features encode normalized (x,y) coordinates. A learned type embedding distinguishes foreground/background points and the two box corners. The same image features are reused when a prompt changes.", "view": { "camera": { "x": 308.5, "y": -144, "z": 1.25 }, "nodePositions": { "input": { "x": 0, "y": 160 }, "flatten@PromptableMaskDecoder#0": { "x": 340, "y": 160 }, "prompt_encoder": { "x": 680, "y": 160 }, "python@PromptableMaskDecoder#0": { "x": 1020, "y": 160 }, "output_tokens": { "x": 1360, "y": 160 }, "cat@PromptableMaskDecoder#0": { "x": 1700, "y": 160 }, "no_mask": { "x": 2040, "y": 160 }, "add@PromptableMaskDecoder#0": { "x": 2380, "y": 160 }, "decoder": { "x": 2720, "y": 160 }, "upscale": { "x": 3060, "y": 160 }, "masks": { "x": 3400, "y": 160 }, "flatten@PromptableMaskDecoder#1": { "x": 3740, "y": 160 }, "quality": { "x": 4080, "y": 160 }, "cat@PromptableMaskDecoder#1": { "x": 4420, "y": 160 }, "output": { "x": 4760, "y": 160 }, "prompt_encoder.position": { "x": 24, "y": 214 }, "prompt_encoder.type_embedding": { "x": 24, "y": 364 }, "add@prompt_encoder#0": { "x": 24, "y": 514 }, "cat@prompt_encoder.position#0": { "x": 24, "y": 214 }, "decoder.first": { "x": 24, "y": 214 }, "decoder.second": { "x": 24, "y": 364 }, "add@decoder#0": { "x": 24, "y": 514 }, "add@decoder#1": { "x": 24, "y": 664 }, "decoder.final_attention": { "x": 24, "y": 814 }, "add@decoder#2": { "x": 24, "y": 964 }, "decoder.final_norm": { "x": 24, "y": 1114 }, "decoder.first.self_attention": { "x": 24, "y": 214 }, "add@decoder.first#0": { "x": 24, "y": 364 }, "decoder.first.self_norm": { "x": 24, "y": 514 }, "add@decoder.first#1": { "x": 24, "y": 664 }, "decoder.first.token_to_image": { "x": 24, "y": 814 }, "add@decoder.first#2": { "x": 24, "y": 964 }, "decoder.first.token_norm": { "x": 24, "y": 1114 }, "decoder.first.up": { "x": 24, "y": 1264 }, "decoder.first.relu": { "x": 24, "y": 1414 }, "decoder.first.down": { "x": 24, "y": 1564 }, "add@decoder.first#3": { "x": 24, "y": 1714 }, "decoder.first.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.first#4": { "x": 24, "y": 2014 }, "decoder.first.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.first#5": { "x": 24, "y": 2314 }, "decoder.first.image_norm": { "x": 24, "y": 2464 }, "decoder.first.self_attention.query": { "x": 24, "y": 214 }, "decoder.first.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.first.self_attention#0": { "x": 24, "y": 514 }, "decoder.first.self_attention.value": { "x": 24, "y": 664 }, "decoder.first.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.first.self_attention.project": { "x": 24, "y": 964 }, "decoder.first.token_to_image.query": { "x": 24, "y": 214 }, "decoder.first.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.first.token_to_image#0": { "x": 24, "y": 514 }, "decoder.first.token_to_image.value": { "x": 24, "y": 664 }, "decoder.first.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.first.token_to_image.project": { "x": 24, "y": 964 }, "decoder.first.image_to_token.query": { "x": 24, "y": 214 }, "decoder.first.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.first.image_to_token#0": { "x": 24, "y": 514 }, "decoder.first.image_to_token.value": { "x": 24, "y": 664 }, "decoder.first.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.first.image_to_token.project": { "x": 24, "y": 964 }, "decoder.second.self_attention": { "x": 24, "y": 214 }, "add@decoder.second#0": { "x": 24, "y": 364 }, "decoder.second.self_norm": { "x": 24, "y": 514 }, "add@decoder.second#1": { "x": 24, "y": 664 }, "decoder.second.token_to_image": { "x": 24, "y": 814 }, "add@decoder.second#2": { "x": 24, "y": 964 }, "decoder.second.token_norm": { "x": 24, "y": 1114 }, "decoder.second.up": { "x": 24, "y": 1264 }, "decoder.second.relu": { "x": 24, "y": 1414 }, "decoder.second.down": { "x": 24, "y": 1564 }, "add@decoder.second#3": { "x": 24, "y": 1714 }, "decoder.second.mlp_norm": { "x": 24, "y": 1864 }, "add@decoder.second#4": { "x": 24, "y": 2014 }, "decoder.second.image_to_token": { "x": 24, "y": 2164 }, "add@decoder.second#5": { "x": 24, "y": 2314 }, "decoder.second.image_norm": { "x": 24, "y": 2464 }, "decoder.second.self_attention.query": { "x": 24, "y": 214 }, "decoder.second.self_attention.key": { "x": 24, "y": 364 }, "div@decoder.second.self_attention#0": { "x": 24, "y": 514 }, "decoder.second.self_attention.value": { "x": 24, "y": 664 }, "decoder.second.self_attention.softmax": { "x": 24, "y": 814 }, "decoder.second.self_attention.project": { "x": 24, "y": 964 }, "decoder.second.token_to_image.query": { "x": 24, "y": 214 }, "decoder.second.token_to_image.key": { "x": 24, "y": 364 }, "div@decoder.second.token_to_image#0": { "x": 24, "y": 514 }, "decoder.second.token_to_image.value": { "x": 24, "y": 664 }, "decoder.second.token_to_image.softmax": { "x": 24, "y": 814 }, "decoder.second.token_to_image.project": { "x": 24, "y": 964 }, "decoder.second.image_to_token.query": { "x": 24, "y": 214 }, "decoder.second.image_to_token.key": { "x": 24, "y": 364 }, "div@decoder.second.image_to_token#0": { "x": 24, "y": 514 }, "decoder.second.image_to_token.value": { "x": 24, "y": 664 }, "decoder.second.image_to_token.softmax": { "x": 24, "y": 814 }, "decoder.second.image_to_token.project": { "x": 24, "y": 964 }, "decoder.final_attention.query": { "x": 24, "y": 214 }, "decoder.final_attention.key": { "x": 24, "y": 364 }, "div@decoder.final_attention#0": { "x": 24, "y": 514 }, "decoder.final_attention.value": { "x": 24, "y": 664 }, "decoder.final_attention.softmax": { "x": 24, "y": 814 }, "decoder.final_attention.project": { "x": 24, "y": 964 }, "upscale.first": { "x": 24, "y": 214 }, "permute@upscale#0": { "x": 24, "y": 364 }, "upscale.norm": { "x": 24, "y": 514 }, "permute@upscale#1": { "x": 24, "y": 664 }, "gelu@upscale#0": { "x": 24, "y": 814 }, "upscale.second": { "x": 24, "y": 964 }, "gelu@upscale#1": { "x": 24, "y": 1114 }, "flatten@masks#0": { "x": 24, "y": 214 }, "masks.mask0": { "x": 24, "y": 364 }, "masks.mask1": { "x": 24, "y": 514 }, "masks.mask2": { "x": 24, "y": 664 }, "stack@masks#0": { "x": 24, "y": 814 }, "python@masks#0": { "x": 24, "y": 964 }, "masks.mask0.first": { "x": 24, "y": 214 }, "relu@masks.mask0#0": { "x": 24, "y": 364 }, "masks.mask0.second": { "x": 24, "y": 514 }, "relu@masks.mask0#1": { "x": 24, "y": 664 }, "masks.mask0.project": { "x": 24, "y": 814 }, "masks.mask1.first": { "x": 24, "y": 214 }, "relu@masks.mask1#0": { "x": 24, "y": 364 }, "masks.mask1.second": { "x": 24, "y": 514 }, "relu@masks.mask1#1": { "x": 24, "y": 664 }, "masks.mask1.project": { "x": 24, "y": 814 }, "masks.mask2.first": { "x": 24, "y": 214 }, "relu@masks.mask2#0": { "x": 24, "y": 364 }, "masks.mask2.second": { "x": 24, "y": 514 }, "relu@masks.mask2#1": { "x": 24, "y": 664 }, "masks.mask2.project": { "x": 24, "y": 814 }, "quality.first": { "x": 24, "y": 214 }, "relu@quality#0": { "x": 24, "y": 364 }, "quality.second": { "x": 24, "y": 514 }, "relu@quality#1": { "x": 24, "y": 664 }, "quality.scores": { "x": 24, "y": 814 } }, "expandedModuleIds": [ "prompt_encoder" ], "selectedNodeIds": [ "prompt_encoder.type_embedding" ], "viewport": { "width": 1100, "height": 700 } }, "anchors": [ { "target": { "kind": "node", "nodeId": "prompt_encoder.type_embedding" }, "snapshotId": "snapshot-2b9a0ff6ffe7a61c", "exportedAt": "2026-09-17T07:00:38.449307+00:00", "modelName": "PromptableMaskDecoder", "context": "{\"connections\":[\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"input\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"prompt_encoder.type_embedding\\\",\\\"targetPort\\\":\\\"x\\\"}\",\"{\\\"kind\\\":\\\"edge\\\",\\\"source\\\":\\\"prompt_encoder.type_embedding\\\",\\\"sourcePort\\\":\\\"y\\\",\\\"target\\\":\\\"add@prompt_encoder#0\\\",\\\"targetPort\\\":\\\"x\\\"}\"],\"contents\":[],\"node\":{\"id\":\"prompt_encoder.type_embedding\",\"kind\":\"layer\",\"op\":\"Embedding\",\"params\":{\"embedding_dim\":8,\"num_embeddings\":4},\"parent\":\"prompt_encoder\"}}" } ] } ], "openingViewId": "prompt" } }