[ { "function_name": "__init__", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "super().__init__", "functionality": "Initialize the parent SASRec class (transformer-based sequential encoder).", "input": "config", "output": "None", "location_or_namespace": "models/SASRec/_model.py" }, { "function_call_name": "nn.Embedding", "functionality": "Creates an embedding lookup table for the 'none' embedding.", "input": "num_embeddings=1, embedding_dim=config['hidden_size']", "output": "Embedding module", "location_or_namespace": "torch.nn" }, { "function_call_name": "nn.init.normal_", "functionality": "Fills the embedding weights with random values from a normal distribution.", "input": "self.none_embedding.weight, mean=0, std=1", "output": "None", "location_or_namespace": "torch.nn.init" }, { "function_call_name": "PreferenceDiffusion", "functionality": "", "input": "config=config", "output": "PreferenceDiffusion object", "location_or_namespace": "diffusion.py" }, { "function_call_name": "SinusoidalPositionEmbeddings", "functionality": "", "input": "config['hidden_size']", "output": "Module", "location_or_namespace": "modules.py" }, { "function_call_name": "nn.Linear", "functionality": "Applies a linear transformation to the incoming data.", "input": "varies (input/output dims)", "output": "Linear module", "location_or_namespace": "torch.nn" }, { "function_call_name": "nn.GELU", "functionality": "Applies the GELU activation function.", "input": "input tensor", "output": "output tensor", "location_or_namespace": "torch.nn" }, { "function_call_name": "nn.Sequential", "functionality": "Stacks modules together in sequence.", "input": "modules", "output": "Sequential module", "location_or_namespace": "torch.nn" } ] }, { "function_name": "get_embeddings", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.item_embeddings", "functionality": "", "input": "items", "output": "item embeddings", "location_or_namespace": "" }, { "function_call_name": "to", "functionality": "Moves tensor to specified device.", "input": "items.device", "output": "Tensor on new device", "location_or_namespace": "torch.Tensor" } ] }, { "function_name": "get_all_embeddings", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.item_embeddings.weight.data", "functionality": "Access all embedding weights.", "input": "", "output": "Tensor", "location_or_namespace": "torch.nn.Embedding" }, { "function_call_name": "to", "functionality": "Moves tensor to specified device.", "input": "device", "output": "Tensor on new device", "location_or_namespace": "torch.Tensor" } ] }, { "function_name": "get_current_embeddings", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.item_embeddings.weight.data", "functionality": "Access all embedding weights.", "input": "", "output": "Tensor", "location_or_namespace": "torch.nn.Embedding" }, { "function_call_name": "to", "functionality": "Moves tensor to specified device.", "input": "device", "output": "Tensor on new device", "location_or_namespace": "torch.Tensor" } ] }, { "function_name": "load_item_embeddings", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "nn.Embedding", "functionality": "Creates an embedding lookup table for items.", "input": "num_embeddings, embedding_dim, padding_idx", "output": "Embedding module", "location_or_namespace": "torch.nn" }, { "function_call_name": "nn.init.normal_", "functionality": "Fills the embedding weights with random values from a normal distribution.", "input": "self.item_embeddings.weight, mean=0, std=1", "output": "None", "location_or_namespace": "torch.nn.init" }, { "function_call_name": "pickle.load", "functionality": "Load object from file.", "input": "open(file_path, 'rb')", "output": "Python object", "location_or_namespace": "pickle" }, { "function_call_name": "torch.tensor", "functionality": "Creates a tensor from data.", "input": "data", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "diagonalize_and_scale", "functionality": "", "input": "tensor", "output": "scaled tensor", "location_or_namespace": "modules.py" }, { "function_call_name": "torch.randn_like", "functionality": "Returns a tensor with the same shape as input filled with random numbers from a normal distribution.", "input": "input_tensor", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "reshape", "functionality": "Reshapes the tensor.", "input": "shape", "output": "reshaped tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "torch.cat", "functionality": "Concatenates tensors along a given dimension.", "input": "tensors, dim", "output": "Tensor", "location_or_namespace": "torch" } ] }, { "function_name": "calcu_h", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "torch.rand", "functionality": "Returns a tensor filled with random numbers from a uniform distribution on [0,1).", "input": "B", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "torch.sign", "functionality": "Returns a new tensor with the signs of the elements of input.", "input": "tensor", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "view", "functionality": "Reshapes the tensor without changing its data.", "input": "B, 1", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "torch.cat", "functionality": "Concatenates tensors along a given dimension.", "input": "tensors, dim", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "to", "functionality": "Moves tensor to specified device.", "input": "logits.device", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "self.none_embedding", "functionality": "", "input": "torch.tensor([0]).to(logits.device)", "output": "embedding", "location_or_namespace": "" } ] }, { "function_name": "denoise_uncon", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.none_embedding", "functionality": "", "input": "torch.tensor([0]).to(x.device)", "output": "embedding", "location_or_namespace": "" }, { "function_call_name": "view", "functionality": "Reshapes the tensor.", "input": "1, x.shape[1]", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "torch.cat", "functionality": "Concatenates tensors along a given dimension.", "input": "tensors, dim", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "self.denoise", "functionality": "", "input": "x, h, step", "output": "", "location_or_namespace": "" } ] }, { "function_name": "denoise", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.step_nn", "functionality": "", "input": "step", "output": "", "location_or_namespace": "" }, { "function_call_name": "torch.cat", "functionality": "Concatenates tensors along a given dimension.", "input": "tensors, dim", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "self.denoise_nn", "functionality": "", "input": "torch.cat((x, h, t), dim=1) OR input", "output": "", "location_or_namespace": "" }, { "function_call_name": "view", "functionality": "Reshapes the tensor.", "input": "B, N, D", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "unsqueeze", "functionality": "Returns a new tensor with a dimension of size one inserted at the specified position.", "input": "1", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "repeat", "functionality": "Repeats the tensor along specified dimensions.", "input": "1, N, 1", "output": "Tensor", "location_or_namespace": "torch.Tensor" } ] }, { "function_name": "forward", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.get_representation", "functionality": "", "input": "batch", "output": "", "location_or_namespace": "" }, { "function_call_name": "self._generate_negative_samples", "functionality": "", "input": "batch", "output": "", "location_or_namespace": "" }, { "function_call_name": "view", "functionality": "Reshapes the tensor.", "input": "-1", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "self.get_embeddings", "functionality": "", "input": "labels", "output": "", "location_or_namespace": "" }, { "function_call_name": "self.get_embeddings", "functionality": "", "input": "labels_neg", "output": "", "location_or_namespace": "" }, { "function_call_name": "self.calcu_h", "functionality": "", "input": "state_hidden, self.config['p']", "output": "", "location_or_namespace": "" }, { "function_call_name": "torch.randint", "functionality": "Returns a tensor with random integers in the specified range.", "input": "0, self.config['timesteps'], (labels.shape[0],), device=h.device", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "self.diff.p_losses", "functionality": "", "input": "self, x_start, x_start_neg, h, n, loss_type=self.config['loss_type']", "output": "loss, _", "location_or_namespace": "" } ] }, { "function_name": "predict", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.get_representation", "functionality": "", "input": "batch", "output": "", "location_or_namespace": "" }, { "function_call_name": "self.diff.sample", "functionality": "", "input": "self, state_hidden", "output": "", "location_or_namespace": "" }, { "function_call_name": "self.get_all_embeddings", "functionality": "", "input": "state_hidden.device", "output": "", "location_or_namespace": "" }, { "function_call_name": "torch.matmul", "functionality": "Performs a matrix multiplication of the two tensors.", "input": "x, test_item_emb.transpose(0, 1)", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "topk", "functionality": "Returns the top k elements and their indices along the specified dimension.", "input": "n_return_sequences, dim=-1", "output": "indices", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "self.get_samples", "functionality": "", "input": "batch", "output": "", "location_or_namespace": "" }, { "function_call_name": "self.get_embeddings", "functionality": "", "input": "torch.tensor([30724]).to(x.device)", "output": "", "location_or_namespace": "" }, { "function_call_name": "self.get_embeddings", "functionality": "", "input": "preds", "output": "", "location_or_namespace": "" } ] }, { "function_name": "get_samples", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "self.get_representation", "functionality": "", "input": "batch", "output": "", "location_or_namespace": "" }, { "function_call_name": "tqdm.tqdm", "functionality": "Display a progress bar for the loop.", "input": "range(1000)", "output": "iterator", "location_or_namespace": "tqdm" }, { "function_call_name": "self.diff.sample", "functionality": "", "input": "self, state_hidden", "output": "", "location_or_namespace": "" }, { "function_call_name": "detach", "functionality": "Returns a new tensor, detached from the current graph.", "input": "", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "cpu", "functionality": "Moves tensor to CPU.", "input": "", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "numpy", "functionality": "Converts a tensor to a NumPy array.", "input": "", "output": "ndarray", "location_or_namespace": "torch.Tensor" } ] }, { "function_name": "_generate_negative_samples", "file_path": "models/PreferDiff.py", "function_calls": [ { "function_call_name": "in_batch_negative_sampling", "functionality": "", "input": "batch['labels']", "output": "", "location_or_namespace": "modules.py" }, { "function_call_name": "in_batch_negative_sampling_sample", "functionality": "", "input": "batch['labels'], self.config['neg_samples']", "output": "", "location_or_namespace": "modules.py" }, { "function_call_name": "np.random.choice", "functionality": "Generates a random sample from a given 1-D array.", "input": "range(self.config['select_pool'][0], self.config['select_pool'][1]), size=1, replace=False", "output": "ndarray", "location_or_namespace": "numpy" }, { "function_call_name": "torch.LongTensor", "functionality": "Constructs a tensor of type long.", "input": "labels_neg", "output": "Tensor", "location_or_namespace": "torch" }, { "function_call_name": "to", "functionality": "Moves tensor to specified device.", "input": "batch['labels'].device", "output": "Tensor", "location_or_namespace": "torch.Tensor" }, { "function_call_name": "reshape", "functionality": "Reshapes the tensor.", "input": "-1, 1", "output": "Tensor", "location_or_namespace": "torch.Tensor" } ] } ]