Failed Cases When Testing with Pytorch v1.12

mert-kurttutan commented 2 years ago

Describe the bug Using current main branch (without any change in the code), several test cases fail

To Reproduce Steps to reproduce the behavior:

Clone the project to your local machine and install required packages (requirements.txt and requirements-dev.txt)
Make sure installed Pytorch is version 1.12 (latest stable version)
Go to root folder (torchinfo) of the project
Run pytest --overwrite (as suggested in README.md)
Observe the failed cases

Expected behavior I think it's supposed to give no failed case (maybe a few warning)

Desktop:

OS: Ubuntu 20.04 LTS

More details After running pytest, summary output from pytest is the following (click for the detailed output):

``` ========================================================================================== short test summary info ========================================================================================== FAILED tests/exceptions_test.py::test_input_size_half_precision - RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [] FAILED tests/torchinfo_test.py::test_pack_padded - RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [Embedding: 1] FAILED tests/torchinfo_test.py::test_namedtuple - RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [] FAILED tests/torchinfo_xl_test.py::test_eval_order_doesnt_matter - RuntimeError: Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same or input should be a MKLDNN te... ============================================================================ 4 failed, 65 passed, 1 skipped, 3 warnings in 8.47s ============================================================================ ```

``` ============================================================================================ test session starts ============================================================================================ platform linux -- Python 3.9.12, pytest-7.1.2, pluggy-1.0.0 rootdir: /home/mertkurttutan/Desktop/main/software-dev/pytest/torchinfo plugins: cov-3.0.0 collected 70 items tests/exceptions_test.py ...F. [ 7%] tests/gpu_test.py .. [ 10%] tests/half_precision_test.py ... [ 14%] tests/torchinfo_test.py ............................F......F.............. [ 85%] tests/torchinfo_xl_test.py ..F...s... [100%] ================================================================================================= FAILURES ================================================================================================== ______________________________________________________________________________________ test_input_size_half_precision _______________________________________________________________________________________ model = Linear(in_features=2, out_features=5, bias=True) x = [tensor([[0.6099, 0.2002], [0.7334, 0.5176], [0.0652, 0.5923], [0.8931, 0.7656], [0.12... [0.9878, 0.7974], [0.8638, 0.2712], [0.3899, 0.2676], [0.9009, 0.7832]], dtype=torch.float16)] batch_dim = None, cache_forward_pass = False, device = 'cpu', mode = , kwargs = {}, model_name = 'Linear', summary_list = [Linear: 0], global_layer_info = {140396508354400: Linear: 0} hooks = {140396508354400: (, )}, saved_model_mode = True def forward_pass( model: nn.Module, x: CORRECTED_INPUT_DATA_TYPE, batch_dim: int | None, cache_forward_pass: bool, device: torch.device | str, mode: Mode, **kwargs: Any, ) -> list[LayerInfo]: """Perform a forward pass on the model using forward hooks.""" global _cached_forward_pass # pylint: disable=global-variable-not-assigned model_name = model.__class__.__name__ if cache_forward_pass and model_name in _cached_forward_pass: return _cached_forward_pass[model_name] summary_list, global_layer_info, hooks = apply_hooks( model_name, model, x, batch_dim ) if x is None: set_children_layers(summary_list) return summary_list kwargs = set_device(kwargs, device) saved_model_mode = model.training try: if mode == Mode.TRAIN: model.train() elif mode == Mode.EVAL: model.eval() else: raise RuntimeError( f"Specified model mode ({list(Mode)}) not recognized: {mode}" ) with torch.no_grad(): # type: ignore[no-untyped-call] if isinstance(x, (list, tuple)): > _ = model.to(device)(*x, **kwargs) torchinfo/torchinfo.py:290: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = Linear(in_features=2, out_features=5, bias=True) input = (tensor([[0.6099, 0.2002], [0.7334, 0.5176], [0.0652, 0.5923], [0.8931, 0.7656], [0.12...[0.9878, 0.7974], [0.8638, 0.2712], [0.3899, 0.2676], [0.9009, 0.7832]], dtype=torch.float16),) kwargs = {}, forward_call = , full_backward_hooks = [], non_full_backward_hooks = [] hook = .pre_hook at 0x7fb0a424ff70>, result = None, bw_hook = None def _call_impl(self, *input, **kwargs): forward_call = (self._slow_forward if torch._C._get_tracing_state() else self.forward) # If we don't have any hooks, we want to skip the rest of the logic in # this function, and just call forward. if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks or _global_forward_hooks or _global_forward_pre_hooks): return forward_call(*input, **kwargs) # Do not call functions when jit is used full_backward_hooks, non_full_backward_hooks = [], [] if self._backward_hooks or _global_backward_hooks: full_backward_hooks, non_full_backward_hooks = self._get_backward_hooks() if _global_forward_pre_hooks or self._forward_pre_hooks: for hook in (*_global_forward_pre_hooks.values(), *self._forward_pre_hooks.values()): result = hook(self, input) if result is not None: if not isinstance(result, tuple): result = (result,) input = result bw_hook = None if full_backward_hooks: bw_hook = hooks.BackwardHook(self, full_backward_hooks) input = bw_hook.setup_input_hook(input) > result = forward_call(*input, **kwargs) ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/module.py:1148: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = Linear(in_features=2, out_features=5, bias=True) input = tensor([[0.6099, 0.2002], [0.7334, 0.5176], [0.0652, 0.5923], [0.8931, 0.7656], [0.123... [0.9878, 0.7974], [0.8638, 0.2712], [0.3899, 0.2676], [0.9009, 0.7832]], dtype=torch.float16) def forward(self, input: Tensor) -> Tensor: > return F.linear(input, self.weight, self.bias) E RuntimeError: "addmm_impl_cpu_" not implemented for 'Half' ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/linear.py:114: RuntimeError The above exception was the direct cause of the following exception: def test_input_size_half_precision() -> None: test = torch.nn.Linear(2, 5).half() with pytest.warns( UserWarning, match=( "Half precision is not supported with input_size parameter, and " "may output incorrect results. Try passing input_data directly." ), ): > summary(test, dtypes=[torch.float16], input_size=(10, 2), device="cpu") tests/exceptions_test.py:59: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ torchinfo/torchinfo.py:218: in summary summary_list = forward_pass( _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ model = Linear(in_features=2, out_features=5, bias=True) x = [tensor([[0.6099, 0.2002], [0.7334, 0.5176], [0.0652, 0.5923], [0.8931, 0.7656], [0.12... [0.9878, 0.7974], [0.8638, 0.2712], [0.3899, 0.2676], [0.9009, 0.7832]], dtype=torch.float16)] batch_dim = None, cache_forward_pass = False, device = 'cpu', mode = , kwargs = {}, model_name = 'Linear', summary_list = [Linear: 0], global_layer_info = {140396508354400: Linear: 0} hooks = {140396508354400: (, )}, saved_model_mode = True def forward_pass( model: nn.Module, x: CORRECTED_INPUT_DATA_TYPE, batch_dim: int | None, cache_forward_pass: bool, device: torch.device | str, mode: Mode, **kwargs: Any, ) -> list[LayerInfo]: """Perform a forward pass on the model using forward hooks.""" global _cached_forward_pass # pylint: disable=global-variable-not-assigned model_name = model.__class__.__name__ if cache_forward_pass and model_name in _cached_forward_pass: return _cached_forward_pass[model_name] summary_list, global_layer_info, hooks = apply_hooks( model_name, model, x, batch_dim ) if x is None: set_children_layers(summary_list) return summary_list kwargs = set_device(kwargs, device) saved_model_mode = model.training try: if mode == Mode.TRAIN: model.train() elif mode == Mode.EVAL: model.eval() else: raise RuntimeError( f"Specified model mode ({list(Mode)}) not recognized: {mode}" ) with torch.no_grad(): # type: ignore[no-untyped-call] if isinstance(x, (list, tuple)): _ = model.to(device)(*x, **kwargs) elif isinstance(x, dict): _ = model.to(device)(**x, **kwargs) else: # Should not reach this point, since process_input_data ensures # x is either a list, tuple, or dict raise ValueError("Unknown input type") except Exception as e: executed_layers = [layer for layer in summary_list if layer.executed] > raise RuntimeError( "Failed to run torchinfo. See above stack traces for more details. " f"Executed layers up to: {executed_layers}" ) from e E RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [] torchinfo/torchinfo.py:299: RuntimeError _____________________________________________________________________________________________ test_pack_padded ______________________________________________________________________________________________ model = PackPaddedLSTM( (embedding): Embedding(60, 128) (lstm): LSTM(128, 32) (hidden2out): Linear(in_features=32, out_features=18, bias=True) (dropout_layer): Dropout(p=0.2, inplace=False) ) x = [tensor([[1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], ..., [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1]], device='cuda:0')] batch_dim = None, cache_forward_pass = False, device = device(type='cuda'), mode = kwargs = {'lengths': tensor([13, 12, 11, 11, 11, 11, 11, 11, 11, 11, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10... 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4], device='cuda:0')} model_name = 'PackPaddedLSTM', summary_list = [PackPaddedLSTM: 0, Embedding: 1] global_layer_info = {140396508397776: Linear: 1, 140396508398688: Embedding: 1, 140396508399792: PackPaddedLSTM: 0, 140396508399840: LSTM: 1, ...} hooks = {140396508397776: (, , ), ...} saved_model_mode = True def forward_pass( model: nn.Module, x: CORRECTED_INPUT_DATA_TYPE, batch_dim: int | None, cache_forward_pass: bool, device: torch.device | str, mode: Mode, **kwargs: Any, ) -> list[LayerInfo]: """Perform a forward pass on the model using forward hooks.""" global _cached_forward_pass # pylint: disable=global-variable-not-assigned model_name = model.__class__.__name__ if cache_forward_pass and model_name in _cached_forward_pass: return _cached_forward_pass[model_name] summary_list, global_layer_info, hooks = apply_hooks( model_name, model, x, batch_dim ) if x is None: set_children_layers(summary_list) return summary_list kwargs = set_device(kwargs, device) saved_model_mode = model.training try: if mode == Mode.TRAIN: model.train() elif mode == Mode.EVAL: model.eval() else: raise RuntimeError( f"Specified model mode ({list(Mode)}) not recognized: {mode}" ) with torch.no_grad(): # type: ignore[no-untyped-call] if isinstance(x, (list, tuple)): > _ = model.to(device)(*x, **kwargs) torchinfo/torchinfo.py:290: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = PackPaddedLSTM( (embedding): Embedding(60, 128) (lstm): LSTM(128, 32) (hidden2out): Linear(in_features=32, out_features=18, bias=True) (dropout_layer): Dropout(p=0.2, inplace=False) ) input = (tensor([[1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], ..., [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1]], device='cuda:0'),) kwargs = {'lengths': tensor([13, 12, 11, 11, 11, 11, 11, 11, 11, 11, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10... 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4], device='cuda:0')} forward_call = full_backward_hooks = [], non_full_backward_hooks = [], hook = .pre_hook at 0x7fb096abd790>, result = None, bw_hook = None def _call_impl(self, *input, **kwargs): forward_call = (self._slow_forward if torch._C._get_tracing_state() else self.forward) # If we don't have any hooks, we want to skip the rest of the logic in # this function, and just call forward. if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks or _global_forward_hooks or _global_forward_pre_hooks): return forward_call(*input, **kwargs) # Do not call functions when jit is used full_backward_hooks, non_full_backward_hooks = [], [] if self._backward_hooks or _global_backward_hooks: full_backward_hooks, non_full_backward_hooks = self._get_backward_hooks() if _global_forward_pre_hooks or self._forward_pre_hooks: for hook in (*_global_forward_pre_hooks.values(), *self._forward_pre_hooks.values()): result = hook(self, input) if result is not None: if not isinstance(result, tuple): result = (result,) input = result bw_hook = None if full_backward_hooks: bw_hook = hooks.BackwardHook(self, full_backward_hooks) input = bw_hook.setup_input_hook(input) > result = forward_call(*input, **kwargs) ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/module.py:1148: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = PackPaddedLSTM( (embedding): Embedding(60, 128) (lstm): LSTM(128, 32) (hidden2out): Linear(in_features=32, out_features=18, bias=True) (dropout_layer): Dropout(p=0.2, inplace=False) ) batch = tensor([[1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], ..., [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1]], device='cuda:0') lengths = tensor([13, 12, 11, 11, 11, 11, 11, 11, 11, 11, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 9, 9..., 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4], device='cuda:0') def forward(self, batch: torch.Tensor, lengths: torch.Tensor) -> torch.Tensor: hidden1 = torch.ones(1, batch.size(-1), self.hidden_size, device=batch.device) hidden2 = torch.ones(1, batch.size(-1), self.hidden_size, device=batch.device) embeds = self.embedding(batch) > packed_input = pack_padded_sequence(embeds, lengths) tests/fixtures/models.py:393: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ input = tensor([[[-1.0181, 0.6983, 1.2245, ..., -0.4112, 0.9594, -0.8608], [-1.0181, 0.6983, 1.2245, ..., -0.4...12, 0.9594, -0.8608], [-1.0181, 0.6983, 1.2245, ..., -0.4112, 0.9594, -0.8608]]], device='cuda:0') lengths = tensor([13, 12, 11, 11, 11, 11, 11, 11, 11, 11, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 9, 9..., 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4], device='cuda:0') batch_first = False, enforce_sorted = True def pack_padded_sequence( input: Tensor, lengths: Tensor, batch_first: bool = False, enforce_sorted: bool = True, ) -> PackedSequence: r"""Packs a Tensor containing padded sequences of variable length. :attr:`input` can be of size ``T x B x *`` where `T` is the length of the longest sequence (equal to ``lengths[0]``), ``B`` is the batch size, and ``*`` is any number of dimensions (including 0). If ``batch_first`` is ``True``, ``B x T x *`` :attr:`input` is expected. For unsorted sequences, use `enforce_sorted = False`. If :attr:`enforce_sorted` is ``True``, the sequences should be sorted by length in a decreasing order, i.e. ``input[:,0]`` should be the longest sequence, and ``input[:,B-1]`` the shortest one. `enforce_sorted = True` is only necessary for ONNX export. Note: This function accepts any input that has at least two dimensions. You can apply it to pack the labels, and use the output of the RNN with them to compute the loss directly. A Tensor can be retrieved from a :class:`PackedSequence` object by accessing its ``.data`` attribute. Args: input (Tensor): padded batch of variable length sequences. lengths (Tensor or list(int)): list of sequence lengths of each batch element (must be on the CPU if provided as a tensor). batch_first (bool, optional): if ``True``, the input is expected in ``B x T x *`` format. enforce_sorted (bool, optional): if ``True``, the input is expected to contain sequences sorted by length in a decreasing order. If ``False``, the input will get sorted unconditionally. Default: ``True``. Returns: a :class:`PackedSequence` object """ if torch._C._get_tracing_state() and not isinstance(lengths, torch.Tensor): warnings.warn('pack_padded_sequence has been called with a Python list of ' 'sequence lengths. The tracer cannot track the data flow of Python ' 'values, and it will treat them as constants, likely rendering ' 'the trace incorrect for any other combination of lengths.', stacklevel=2) lengths = torch.as_tensor(lengths, dtype=torch.int64) if enforce_sorted: sorted_indices = None else: lengths, sorted_indices = torch.sort(lengths, descending=True) sorted_indices = sorted_indices.to(input.device) batch_dim = 0 if batch_first else 1 input = input.index_select(batch_dim, sorted_indices) data, batch_sizes = \ > _VF._pack_padded_sequence(input, lengths, batch_first) E RuntimeError: 'lengths' argument should be a 1D CPU int64 tensor, but got 1D cuda:0 Long tensor ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/utils/rnn.py:260: RuntimeError The above exception was the direct cause of the following exception: def test_pack_padded() -> None: x = torch.ones([20, 128]).long() # fmt: off y = torch.Tensor([ 13, 12, 11, 11, 11, 11, 11, 11, 11, 11, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 8, 8, 8, 8, 8, 8, 8, 8, 8, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4 ]).long() # fmt: on > summary(PackPaddedLSTM(), input_data=x, lengths=y) tests/torchinfo_test.py:307: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ torchinfo/torchinfo.py:218: in summary summary_list = forward_pass( _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ model = PackPaddedLSTM( (embedding): Embedding(60, 128) (lstm): LSTM(128, 32) (hidden2out): Linear(in_features=32, out_features=18, bias=True) (dropout_layer): Dropout(p=0.2, inplace=False) ) x = [tensor([[1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], ..., [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1], [1, 1, 1, ..., 1, 1, 1]], device='cuda:0')] batch_dim = None, cache_forward_pass = False, device = device(type='cuda'), mode = kwargs = {'lengths': tensor([13, 12, 11, 11, 11, 11, 11, 11, 11, 11, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10... 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4], device='cuda:0')} model_name = 'PackPaddedLSTM', summary_list = [PackPaddedLSTM: 0, Embedding: 1] global_layer_info = {140396508397776: Linear: 1, 140396508398688: Embedding: 1, 140396508399792: PackPaddedLSTM: 0, 140396508399840: LSTM: 1, ...} hooks = {140396508397776: (, , ), ...} saved_model_mode = True def forward_pass( model: nn.Module, x: CORRECTED_INPUT_DATA_TYPE, batch_dim: int | None, cache_forward_pass: bool, device: torch.device | str, mode: Mode, **kwargs: Any, ) -> list[LayerInfo]: """Perform a forward pass on the model using forward hooks.""" global _cached_forward_pass # pylint: disable=global-variable-not-assigned model_name = model.__class__.__name__ if cache_forward_pass and model_name in _cached_forward_pass: return _cached_forward_pass[model_name] summary_list, global_layer_info, hooks = apply_hooks( model_name, model, x, batch_dim ) if x is None: set_children_layers(summary_list) return summary_list kwargs = set_device(kwargs, device) saved_model_mode = model.training try: if mode == Mode.TRAIN: model.train() elif mode == Mode.EVAL: model.eval() else: raise RuntimeError( f"Specified model mode ({list(Mode)}) not recognized: {mode}" ) with torch.no_grad(): # type: ignore[no-untyped-call] if isinstance(x, (list, tuple)): _ = model.to(device)(*x, **kwargs) elif isinstance(x, dict): _ = model.to(device)(**x, **kwargs) else: # Should not reach this point, since process_input_data ensures # x is either a list, tuple, or dict raise ValueError("Unknown input type") except Exception as e: executed_layers = [layer for layer in summary_list if layer.executed] > raise RuntimeError( "Failed to run torchinfo. See above stack traces for more details. " f"Executed layers up to: {executed_layers}" ) from e E RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [Embedding: 1] torchinfo/torchinfo.py:299: RuntimeError ______________________________________________________________________________________________ test_namedtuple ______________________________________________________________________________________________ model = NamedTuple() x = [tensor([[[[0.0426, 0.8419, 0.7895, ..., 0.2941, 0.8007, 0.9451], [0.6184, 0.2044, 0.7810, ..., 0.3096, 0.... [5.4683e-01, 3.8055e-01, 9.6086e-01, ..., 1.3338e-01, 6.0339e-01, 7.6829e-01]]]], device='cuda:0')] batch_dim = None, cache_forward_pass = False, device = device(type='cuda'), mode = , kwargs = {'z': Point(x=(2, 1, 28, 28), y=(2, 1, 28, 28))}, model_name = 'NamedTuple' summary_list = [NamedTuple: 0], global_layer_info = {140396508353824: NamedTuple: 0} hooks = {140396508353824: (, )}, saved_model_mode = True def forward_pass( model: nn.Module, x: CORRECTED_INPUT_DATA_TYPE, batch_dim: int | None, cache_forward_pass: bool, device: torch.device | str, mode: Mode, **kwargs: Any, ) -> list[LayerInfo]: """Perform a forward pass on the model using forward hooks.""" global _cached_forward_pass # pylint: disable=global-variable-not-assigned model_name = model.__class__.__name__ if cache_forward_pass and model_name in _cached_forward_pass: return _cached_forward_pass[model_name] summary_list, global_layer_info, hooks = apply_hooks( model_name, model, x, batch_dim ) if x is None: set_children_layers(summary_list) return summary_list kwargs = set_device(kwargs, device) saved_model_mode = model.training try: if mode == Mode.TRAIN: model.train() elif mode == Mode.EVAL: model.eval() else: raise RuntimeError( f"Specified model mode ({list(Mode)}) not recognized: {mode}" ) with torch.no_grad(): # type: ignore[no-untyped-call] if isinstance(x, (list, tuple)): > _ = model.to(device)(*x, **kwargs) torchinfo/torchinfo.py:290: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = NamedTuple() input = (tensor([[[[0.0426, 0.8419, 0.7895, ..., 0.2941, 0.8007, 0.9451], [0.6184, 0.2044, 0.7810, ..., 0.3096, 0.... [5.4683e-01, 3.8055e-01, 9.6086e-01, ..., 1.3338e-01, 6.0339e-01, 7.6829e-01]]]], device='cuda:0')) kwargs = {'z': Point(x=(2, 1, 28, 28), y=(2, 1, 28, 28))}, forward_call = , full_backward_hooks = [], non_full_backward_hooks = [] hook = .pre_hook at 0x7fb096abd940>, result = None, bw_hook = None def _call_impl(self, *input, **kwargs): forward_call = (self._slow_forward if torch._C._get_tracing_state() else self.forward) # If we don't have any hooks, we want to skip the rest of the logic in # this function, and just call forward. if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks or _global_forward_hooks or _global_forward_pre_hooks): return forward_call(*input, **kwargs) # Do not call functions when jit is used full_backward_hooks, non_full_backward_hooks = [], [] if self._backward_hooks or _global_backward_hooks: full_backward_hooks, non_full_backward_hooks = self._get_backward_hooks() if _global_forward_pre_hooks or self._forward_pre_hooks: for hook in (*_global_forward_pre_hooks.values(), *self._forward_pre_hooks.values()): result = hook(self, input) if result is not None: if not isinstance(result, tuple): result = (result,) input = result bw_hook = None if full_backward_hooks: bw_hook = hooks.BackwardHook(self, full_backward_hooks) input = bw_hook.setup_input_hook(input) > result = forward_call(*input, **kwargs) ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/module.py:1148: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = NamedTuple() x = tensor([[[[0.0426, 0.8419, 0.7895, ..., 0.2941, 0.8007, 0.9451], [0.6184, 0.2044, 0.7810, ..., 0.3096, 0.0..., 0.3716, 0.1156, 0.9332], [0.6786, 0.6320, 0.4887, ..., 0.8578, 0.7893, 0.7768]]]], device='cuda:0') y = tensor([[[[8.8717e-02, 3.5092e-01, 3.9958e-01, ..., 9.2321e-01, 1.7570e-01, 9.7187e-01], [9.9338... [5.4683e-01, 3.8055e-01, 9.6086e-01, ..., 1.3338e-01, 6.0339e-01, 7.6829e-01]]]], device='cuda:0') z = Point(x=(2, 1, 28, 28), y=(2, 1, 28, 28)) def forward(self, x: Any, y: Any, z: Any) -> Any: > return self.Point(x, y).x + torch.ones(z.x) E RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! tests/fixtures/models.py:332: RuntimeError The above exception was the direct cause of the following exception: def test_namedtuple() -> None: model = NamedTuple() input_size = [(2, 1, 28, 28), (2, 1, 28, 28)] named_tuple = model.Point(*input_size) > summary(model, input_size=input_size, z=named_tuple) tests/torchinfo_test.py:373: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ torchinfo/torchinfo.py:218: in summary summary_list = forward_pass( _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ model = NamedTuple() x = [tensor([[[[0.0426, 0.8419, 0.7895, ..., 0.2941, 0.8007, 0.9451], [0.6184, 0.2044, 0.7810, ..., 0.3096, 0.... [5.4683e-01, 3.8055e-01, 9.6086e-01, ..., 1.3338e-01, 6.0339e-01, 7.6829e-01]]]], device='cuda:0')] batch_dim = None, cache_forward_pass = False, device = device(type='cuda'), mode = , kwargs = {'z': Point(x=(2, 1, 28, 28), y=(2, 1, 28, 28))}, model_name = 'NamedTuple' summary_list = [NamedTuple: 0], global_layer_info = {140396508353824: NamedTuple: 0} hooks = {140396508353824: (, )}, saved_model_mode = True def forward_pass( model: nn.Module, x: CORRECTED_INPUT_DATA_TYPE, batch_dim: int | None, cache_forward_pass: bool, device: torch.device | str, mode: Mode, **kwargs: Any, ) -> list[LayerInfo]: """Perform a forward pass on the model using forward hooks.""" global _cached_forward_pass # pylint: disable=global-variable-not-assigned model_name = model.__class__.__name__ if cache_forward_pass and model_name in _cached_forward_pass: return _cached_forward_pass[model_name] summary_list, global_layer_info, hooks = apply_hooks( model_name, model, x, batch_dim ) if x is None: set_children_layers(summary_list) return summary_list kwargs = set_device(kwargs, device) saved_model_mode = model.training try: if mode == Mode.TRAIN: model.train() elif mode == Mode.EVAL: model.eval() else: raise RuntimeError( f"Specified model mode ({list(Mode)}) not recognized: {mode}" ) with torch.no_grad(): # type: ignore[no-untyped-call] if isinstance(x, (list, tuple)): _ = model.to(device)(*x, **kwargs) elif isinstance(x, dict): _ = model.to(device)(**x, **kwargs) else: # Should not reach this point, since process_input_data ensures # x is either a list, tuple, or dict raise ValueError("Unknown input type") except Exception as e: executed_layers = [layer for layer in summary_list if layer.executed] > raise RuntimeError( "Failed to run torchinfo. See above stack traces for more details. " f"Executed layers up to: {executed_layers}" ) from e E RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [] torchinfo/torchinfo.py:299: RuntimeError _______________________________________________________________________________________ test_eval_order_doesnt_matter _______________________________________________________________________________________ def test_eval_order_doesnt_matter() -> None: input_size = (1, 3, 224, 224) input_tensor = torch.ones(input_size) model1 = torchvision.models.resnet18(pretrained=True) model1.eval() summary(model1, input_size=input_size) with torch.inference_mode(): # type: ignore[no-untyped-call] > output1 = model1(input_tensor) tests/torchinfo_xl_test.py:43: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/module.py:1130: in _call_impl return forward_call(*input, **kwargs) ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torchvision/models/resnet.py:285: in forward return self._forward_impl(x) ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torchvision/models/resnet.py:268: in _forward_impl x = self.conv1(x) ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/module.py:1130: in _call_impl return forward_call(*input, **kwargs) ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/conv.py:457: in forward return self._conv_forward(input, self.weight, self.bias) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ self = Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False) input = tensor([[[[1., 1., 1., ..., 1., 1., 1.], [1., 1., 1., ..., 1., 1., 1.], [1., 1., 1., ..., 1., 1... [1., 1., 1., ..., 1., 1., 1.], [1., 1., 1., ..., 1., 1., 1.], [1., 1., 1., ..., 1., 1., 1.]]]]) weight = Parameter containing: tensor([[[[-1.0419e-02, -6.1356e-03, -1.8098e-03, ..., 5.6615e-02, 1.7083e-02, -1....0065e-03, 3.6341e-02, ..., -2.4361e-02, -7.1195e-02, -6.6788e-02]]]], device='cuda:0', requires_grad=True) bias = None def _conv_forward(self, input: Tensor, weight: Tensor, bias: Optional[Tensor]): if self.padding_mode != 'zeros': return F.conv2d(F.pad(input, self._reversed_padding_repeated_twice, mode=self.padding_mode), weight, bias, self.stride, _pair(0), self.dilation, self.groups) > return F.conv2d(input, weight, bias, self.stride, self.padding, self.dilation, self.groups) E RuntimeError: Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same or input should be a MKLDNN tensor and weight is a dense tensor ../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/conv.py:453: RuntimeError ------------------------------------------------------------------------------------------- Captured stdout call -------------------------------------------------------------------------------------------- ========================================================================================== Layer (type:depth-idx) Output Shape Param # ========================================================================================== ResNet [1, 1000] -- ├─Conv2d: 1-1 [1, 64, 112, 112] 9,408 ├─BatchNorm2d: 1-2 [1, 64, 112, 112] 128 ├─ReLU: 1-3 [1, 64, 112, 112] -- ├─MaxPool2d: 1-4 [1, 64, 56, 56] -- ├─Sequential: 1-5 [1, 64, 56, 56] -- │ └─BasicBlock: 2-1 [1, 64, 56, 56] -- │ │ └─Conv2d: 3-1 [1, 64, 56, 56] 36,864 │ │ └─BatchNorm2d: 3-2 [1, 64, 56, 56] 128 │ │ └─ReLU: 3-3 [1, 64, 56, 56] -- │ │ └─Conv2d: 3-4 [1, 64, 56, 56] 36,864 │ │ └─BatchNorm2d: 3-5 [1, 64, 56, 56] 128 │ │ └─ReLU: 3-6 [1, 64, 56, 56] -- │ └─BasicBlock: 2-2 [1, 64, 56, 56] -- │ │ └─Conv2d: 3-7 [1, 64, 56, 56] 36,864 │ │ └─BatchNorm2d: 3-8 [1, 64, 56, 56] 128 │ │ └─ReLU: 3-9 [1, 64, 56, 56] -- │ │ └─Conv2d: 3-10 [1, 64, 56, 56] 36,864 │ │ └─BatchNorm2d: 3-11 [1, 64, 56, 56] 128 │ │ └─ReLU: 3-12 [1, 64, 56, 56] -- ├─Sequential: 1-6 [1, 128, 28, 28] -- │ └─BasicBlock: 2-3 [1, 128, 28, 28] -- │ │ └─Conv2d: 3-13 [1, 128, 28, 28] 73,728 │ │ └─BatchNorm2d: 3-14 [1, 128, 28, 28] 256 │ │ └─ReLU: 3-15 [1, 128, 28, 28] -- │ │ └─Conv2d: 3-16 [1, 128, 28, 28] 147,456 │ │ └─BatchNorm2d: 3-17 [1, 128, 28, 28] 256 │ │ └─Sequential: 3-18 [1, 128, 28, 28] 8,448 │ │ └─ReLU: 3-19 [1, 128, 28, 28] -- │ └─BasicBlock: 2-4 [1, 128, 28, 28] -- │ │ └─Conv2d: 3-20 [1, 128, 28, 28] 147,456 │ │ └─BatchNorm2d: 3-21 [1, 128, 28, 28] 256 │ │ └─ReLU: 3-22 [1, 128, 28, 28] -- │ │ └─Conv2d: 3-23 [1, 128, 28, 28] 147,456 │ │ └─BatchNorm2d: 3-24 [1, 128, 28, 28] 256 │ │ └─ReLU: 3-25 [1, 128, 28, 28] -- ├─Sequential: 1-7 [1, 256, 14, 14] -- │ └─BasicBlock: 2-5 [1, 256, 14, 14] -- │ │ └─Conv2d: 3-26 [1, 256, 14, 14] 294,912 │ │ └─BatchNorm2d: 3-27 [1, 256, 14, 14] 512 │ │ └─ReLU: 3-28 [1, 256, 14, 14] -- │ │ └─Conv2d: 3-29 [1, 256, 14, 14] 589,824 │ │ └─BatchNorm2d: 3-30 [1, 256, 14, 14] 512 │ │ └─Sequential: 3-31 [1, 256, 14, 14] 33,280 │ │ └─ReLU: 3-32 [1, 256, 14, 14] -- │ └─BasicBlock: 2-6 [1, 256, 14, 14] -- │ │ └─Conv2d: 3-33 [1, 256, 14, 14] 589,824 │ │ └─BatchNorm2d: 3-34 [1, 256, 14, 14] 512 │ │ └─ReLU: 3-35 [1, 256, 14, 14] -- │ │ └─Conv2d: 3-36 [1, 256, 14, 14] 589,824 │ │ └─BatchNorm2d: 3-37 [1, 256, 14, 14] 512 │ │ └─ReLU: 3-38 [1, 256, 14, 14] -- ├─Sequential: 1-8 [1, 512, 7, 7] -- │ └─BasicBlock: 2-7 [1, 512, 7, 7] -- │ │ └─Conv2d: 3-39 [1, 512, 7, 7] 1,179,648 │ │ └─BatchNorm2d: 3-40 [1, 512, 7, 7] 1,024 │ │ └─ReLU: 3-41 [1, 512, 7, 7] -- │ │ └─Conv2d: 3-42 [1, 512, 7, 7] 2,359,296 │ │ └─BatchNorm2d: 3-43 [1, 512, 7, 7] 1,024 │ │ └─Sequential: 3-44 [1, 512, 7, 7] 132,096 │ │ └─ReLU: 3-45 [1, 512, 7, 7] -- │ └─BasicBlock: 2-8 [1, 512, 7, 7] -- │ │ └─Conv2d: 3-46 [1, 512, 7, 7] 2,359,296 │ │ └─BatchNorm2d: 3-47 [1, 512, 7, 7] 1,024 │ │ └─ReLU: 3-48 [1, 512, 7, 7] -- │ │ └─Conv2d: 3-49 [1, 512, 7, 7] 2,359,296 │ │ └─BatchNorm2d: 3-50 [1, 512, 7, 7] 1,024 │ │ └─ReLU: 3-51 [1, 512, 7, 7] -- ├─AdaptiveAvgPool2d: 1-9 [1, 512, 1, 1] -- ├─Linear: 1-10 [1, 1000] 513,000 ========================================================================================== Total params: 11,689,512 Trainable params: 11,689,512 Non-trainable params: 0 Total mult-adds (G): 1.81 ========================================================================================== Input size (MB): 0.60 Forward/backward pass size (MB): 39.75 Params size (MB): 46.76 Estimated Total Size (MB): 87.11 ========================================================================================== ============================================================================================= warnings summary ============================================================================================== tests/torchinfo_xl_test.py::test_eval_order_doesnt_matter /home/mertkurttutan/miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead. warnings.warn( tests/torchinfo_xl_test.py::test_eval_order_doesnt_matter /home/mertkurttutan/miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights. warnings.warn(msg) tests/torchinfo_xl_test.py::test_google /home/mertkurttutan/miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torchvision/models/googlenet.py:47: FutureWarning: The default weight initialization of GoogleNet will be changed in future releases of torchvision. If you wish to keep the old behavior (which leads to long initialization times due to scipy/scipy#11299), please set init_weights=True. warnings.warn( -- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html ========================================================================================== short test summary info ========================================================================================== FAILED tests/exceptions_test.py::test_input_size_half_precision - RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [] FAILED tests/torchinfo_test.py::test_pack_padded - RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [Embedding: 1] FAILED tests/torchinfo_test.py::test_namedtuple - RuntimeError: Failed to run torchinfo. See above stack traces for more details. Executed layers up to: [] FAILED tests/torchinfo_xl_test.py::test_eval_order_doesnt_matter - RuntimeError: Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same or input should be a MKLDNN te... ============================================================================ 4 failed, 65 passed, 1 skipped, 3 warnings in 8.47s ============================================================================ ```

3 warnings are not that important since they were just deprecation warnings from torchvision. They were resolved once I used the new input formats.

Regarding failed cases, the first one stems from the following runtime error:

self = Linear(in_features=2, out_features=5, bias=True)
input = tensor([[0.6099, 0.2002],
        [0.7334, 0.5176],
        [0.0652, 0.5923],
        [0.8931, 0.7656],
        [0.123...  [0.9878, 0.7974],
        [0.8638, 0.2712],
        [0.3899, 0.2676],
        [0.9009, 0.7832]], dtype=torch.float16)

    def forward(self, input: Tensor) -> Tensor:
>       return F.linear(input, self.weight, self.bias)
E       RuntimeError: "addmm_impl_cpu_" not implemented for 'Half'

../../../../../miniconda3/envs/cs224n_a3/lib/python3.9/site-packages/torch/nn/modules/linear.py:114: RuntimeError

It seems that in Pytorch v1.12, half precision is not supported on cpu ??? (Also see this remark here)

The other 3 failed cases were because inputs tensors are initialized outside summary and with no explicit device. So, they were created on cpu (in my case). But, the model created inside summary uses GPU automatically unless I make cuda unavailable because of the following line in torchinfo.py (summary function)

if device is None:
        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

Indeed, these 3 cases were resolved once I used device="cpu" when running summary.

TylerYep commented 2 years ago

Thank you for looking into this! I would really appreciate a PR fixing these.

I think it should be fine to set that half precision test to run on GPU-only by moving the test to gpu_test.py.

For the other cases, we currently do not run these tests on GPUs (I don't think Github Actions supports this yet), but changes to support these tests passing on both CPU and GPU are welcome

mert-kurttutan commented 2 years ago

I think we should move the first part of test_input_size_half_precision to gpu_test since the case is only about usage of input_size parameter and half.precision.

I am not sure about the second warning case since it is there to raise warning when half precision is used on CPU. Besides this issue, I am ready to submit PR.

TylerYep commented 2 years ago

Yep, that's what I meant

mert-kurttutan commented 2 years ago

Then, I am moving only the first warning case to gpu_test and delete the second warning case since it raises already runtime error, right?

By second warning case, I mean the following in test_input_size_half_precision() :

    with pytest.warns(
        UserWarning,
        match=(
            "Half precision is not supported on cpu. Set the `device` field or "
            "pass `input_data` using the correct device."
        ),
    ):
        summary(
            test,
            dtypes=[torch.float16],
            input_data=torch.randn((10, 2), dtype=torch.float16, device="cpu"),
            device="cpu",
        )

TylerYep commented 2 years ago

Yep, we can remove that test case. We can leave the warning in code though, since it will warn users using earlier versions of PyTorch that do not raise a runtime error

TylerYep / torchinfo

Failed Cases When Testing with Pytorch v1.12 #161