File size: 14,839 Bytes
e89a301 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 | [
{
"function_name": "ContextGNN.__init__",
"file_path": "contextgnn/nn/models/contextgnn.py",
"function_calls": [
{
"function_call_name": "super().__init__()",
"functionality": "Initialize parent class (RHSEmbeddingGNN), which sets up global item embedding and config.",
"input": "Arguments for the parent class RHSEmbeddingGNN.",
"output": "None (initializes attributes and internal states).",
"location_or_namespace": "contextgnn.nn.models.rhsembeddinggnn.RHSEmbeddingGNN"
},
{
"function_call_name": "HeteroEncoder()",
"functionality": "Encodes multi-modal tabular features for each node type.",
"input": "channels, node_to_col_names_dict, node_to_col_stats, stype_encoder_cls_kwargs, torch_frame_model_cls, torch_frame_model_kwargs",
"output": "torch.nn.Module that outputs node features.",
"location_or_namespace": "contextgnn.nn.encoder.HeteroEncoder"
},
{
"function_call_name": "HeteroTemporalEncoder()",
"functionality": "Encodes relative time for each node type as features using positional encodings.",
"input": "node_types, channels",
"output": "torch.nn.Module that outputs temporal node features.",
"location_or_namespace": "contextgnn.nn.encoder.HeteroTemporalEncoder"
},
{
"function_call_name": "HeteroGraphSAGE()",
"functionality": "Heterogeneous GNN for message passing over multi-relation graphs.",
"input": "node_types, edge_types, channels, aggr, num_layers",
"output": "torch.nn.Module that computes updated node features.",
"location_or_namespace": "contextgnn.nn.models.HeteroGraphSAGE"
},
{
"function_call_name": "MLP()",
"functionality": "Multi-layer perceptron for transforming embeddings (used for scoring/prediction).",
"input": "channels, out_channels=1, norm, num_layers",
"output": "MLP module (torch.nn.Module) with output dimension out_channels.",
"location_or_namespace": "torch_geometric.nn.MLP"
},
{
"function_call_name": "torch.nn.Linear()",
"functionality": "Linear transformation (fully connected layer) used for projection and offset scoring.",
"input": "in_features, out_features",
"output": "torch.nn.Linear module.",
"location_or_namespace": "torch.nn"
},
{
"function_call_name": "torch.nn.Embedding()",
"functionality": "Creates an embedding matrix (ID awareness feature for user node).",
"input": "num_embeddings, embedding_dim",
"output": "Embedding lookup module.",
"location_or_namespace": "torch.nn"
}
]
},
{
"function_name": "ContextGNN.reset_parameters",
"file_path": "contextgnn/nn/models/contextgnn.py",
"function_calls": [
{
"function_call_name": "super().reset_parameters()",
"functionality": "Calls reset_parameters on the parent RHSEmbeddingGNN.",
"input": "None",
"output": "None",
"location_or_namespace": "contextgnn.nn.models.rhsembeddinggnn.RHSEmbeddingGNN"
},
{
"function_call_name": "self.encoder.reset_parameters()",
"functionality": "Resets the parameters of the HeteroEncoder module.",
"input": "None",
"output": "None",
"location_or_namespace": "contextgnn.nn.encoder.HeteroEncoder"
},
{
"function_call_name": "self.temporal_encoder.reset_parameters()",
"functionality": "Resets the parameters of the HeteroTemporalEncoder module.",
"input": "None",
"output": "None",
"location_or_namespace": "contextgnn.nn.encoder.HeteroTemporalEncoder"
},
{
"function_call_name": "self.gnn.reset_parameters()",
"functionality": "Resets the parameters of the HeteroGraphSAGE module.",
"input": "None",
"output": "None",
"location_or_namespace": "contextgnn.nn.models.HeteroGraphSAGE"
},
{
"function_call_name": "self.head.reset_parameters()",
"functionality": "Resets the parameters of the MLP head.",
"input": "None",
"output": "None",
"location_or_namespace": "torch_geometric.nn.MLP"
},
{
"function_call_name": "self.id_awareness_emb.reset_parameters()",
"functionality": "Resets the parameters of the ID-awareness embedding.",
"input": "None",
"output": "None",
"location_or_namespace": "torch.nn.Embedding"
},
{
"function_call_name": "self.rhs_embedding.reset_parameters()",
"functionality": "Resets the parameters of the right-hand-side (item) shallow embedding matrix.",
"input": "None",
"output": "None",
"location_or_namespace": "contextgnn.nn.models.rhsembeddinggnn.RHSEmbeddingGNN"
},
{
"function_call_name": "self.lin_offset_embgnn.reset_parameters()",
"functionality": "Resets parameters of the embgnn offset linear layer.",
"input": "None",
"output": "None",
"location_or_namespace": "torch.nn.Linear"
},
{
"function_call_name": "self.lin_offset_idgnn.reset_parameters()",
"functionality": "Resets parameters of the idgnn offset linear layer.",
"input": "None",
"output": "None",
"location_or_namespace": "torch.nn.Linear"
},
{
"function_call_name": "self.lhs_projector.reset_parameters()",
"functionality": "Resets parameters of the user embedding projection layer.",
"input": "None",
"output": "None",
"location_or_namespace": "torch.nn.Linear"
}
]
},
{
"function_name": "ContextGNN.sample_step",
"file_path": "contextgnn/nn/models/contextgnn.py",
"function_calls": [
{
"function_call_name": "torch.rand()",
"functionality": "Generates a tensor of random numbers in [0,1).",
"input": "size (int or tuple), device",
"output": "Tensor of random floats.",
"location_or_namespace": "torch"
},
{
"function_call_name": "map_index()",
"functionality": "Remaps indices according to a reference index set, used for advanced indexing/sampling.",
"input": "index, reference index, max_index, inclusive",
"output": "Remapped indices and boolean mask.",
"location_or_namespace": "torch_geometric.utils.map"
},
{
"function_call_name": "self.rhs_embedding()",
"functionality": "Returns the embedding vectors of the sampled item (RHS) indices.",
"input": "rhs_index (Tensor)",
"output": "Tensor of embedding vectors.",
"location_or_namespace": "contextgnn.nn.models.rhsembeddinggnn.RHSEmbeddingGNN"
}
]
},
{
"function_name": "ContextGNN.construct_logits",
"file_path": "contextgnn/nn/models/contextgnn.py",
"function_calls": [
{
"function_call_name": "Tensor.t()",
"functionality": "Returns the transpose of a tensor (for dot product scoring).",
"input": "None",
"output": "Transposed tensor.",
"location_or_namespace": "torch.Tensor"
},
{
"function_call_name": "self.head()",
"functionality": "Applies MLP head to the right-hand-side (item) GNN embeddings for scoring.",
"input": "rhs_gnn_embedding (Tensor)",
"output": "Score tensor.",
"location_or_namespace": "torch_geometric.nn.MLP"
},
{
"function_call_name": "Tensor.flatten()",
"functionality": "Returns a contiguous flattened tensor.",
"input": "start_dim=0, end_dim=-1 (optional)",
"output": "Flattened tensor.",
"location_or_namespace": "torch.Tensor"
},
{
"function_call_name": "Tensor.sum()",
"functionality": "Returns the sum of each row (used for user-item elementwise product).",
"input": "dim (int), keepdim (bool)",
"output": "Summed tensor.",
"location_or_namespace": "torch.Tensor"
}
]
},
{
"function_name": "ContextGNN.forward_gnn",
"file_path": "contextgnn/nn/models/contextgnn.py",
"function_calls": [
{
"function_call_name": "self.encoder()",
"functionality": "Encodes tabular/multimodal node features to embedding tensors.",
"input": "tf_dict (Dict of TensorFrames)",
"output": "Dict of node embeddings.",
"location_or_namespace": "contextgnn.nn.encoder.HeteroEncoder"
},
{
"function_call_name": "self.id_awareness_emb.weight",
"functionality": "Returns the weights of the ID-awareness embedding for addition to node features.",
"input": "None",
"output": "Tensor of embedding weights.",
"location_or_namespace": "torch.nn.Embedding"
},
{
"function_call_name": "self.temporal_encoder()",
"functionality": "Encodes relative time into features for each node type.",
"input": "seed_time (Tensor), time_dict (Dict), batch_dict (Dict)",
"output": "Dict of temporal feature tensors.",
"location_or_namespace": "contextgnn.nn.encoder.HeteroTemporalEncoder"
},
{
"function_call_name": "self.gnn()",
"functionality": "Performs message passing and updates node features with GNN.",
"input": "x_dict (Dict), edge_index_dict (Dict)",
"output": "Dict of updated node embeddings.",
"location_or_namespace": "contextgnn.nn.models.HeteroGraphSAGE"
}
]
},
{
"function_name": "ContextGNN.forward",
"file_path": "contextgnn/nn/models/contextgnn.py",
"function_calls": [
{
"function_call_name": "self.forward_gnn()",
"functionality": "Obtains context-aware embeddings from the GNN.",
"input": "batch (HeteroData), entity_table (NodeType)",
"output": "Dict of node embeddings.",
"location_or_namespace": "self (ContextGNN)"
},
{
"function_call_name": "self.lhs_projector()",
"functionality": "Projects user embeddings into a common space for dot-product scoring.",
"input": "lhs_embedding (Tensor)",
"output": "Projected user embedding (Tensor).",
"location_or_namespace": "torch.nn.Linear"
},
{
"function_call_name": "self.rhs_embedding()",
"functionality": "Returns the embedding vectors of all candidate items (RHS).",
"input": "None",
"output": "Tensor of embedding vectors.",
"location_or_namespace": "contextgnn.nn.models.rhsembeddinggnn.RHSEmbeddingGNN"
},
{
"function_call_name": "self.construct_logits()",
"functionality": "Fuses two-tower and pair-wise logits as final prediction.",
"input": "Various embedding tensors and batch indices.",
"output": "Tensor of fused logits.",
"location_or_namespace": "self (ContextGNN)"
}
]
},
{
"function_name": "ContextGNN.forward_sample_softmax",
"file_path": "contextgnn/nn/models/contextgnn.py",
"function_calls": [
{
"function_call_name": "self.forward_gnn()",
"functionality": "Obtains context-aware embeddings from the GNN.",
"input": "batch (HeteroData), entity_table (NodeType)",
"output": "Dict of node embeddings.",
"location_or_namespace": "self (ContextGNN)"
},
{
"function_call_name": "self.lhs_projector()",
"functionality": "Projects user embeddings into a common space for dot-product scoring.",
"input": "lhs_embedding (Tensor)",
"output": "Projected user embedding (Tensor).",
"location_or_namespace": "torch.nn.Linear"
},
{
"function_call_name": "self.sample_step()",
"functionality": "Selects a subset of RHS indices and associated features for sampled softmax.",
"input": "rhs_idgnn_index, lhs_idgnn_batch, rhs_gnn_embedding, src_batch, dst_index",
"output": "Sampled indices, embeddings, and associated batch indices.",
"location_or_namespace": "self (ContextGNN)"
},
{
"function_call_name": "self.construct_logits()",
"functionality": "Fuses two-tower and pair-wise logits as final prediction.",
"input": "Various embedding tensors and batch indices.",
"output": "Tensor of fused logits.",
"location_or_namespace": "self (ContextGNN)"
}
]
}
]
|