Created
December 17, 2022 19:09
-
-
Save ezyang/a1b15c758693ef52979676f1cb015667 to your computer and use it in GitHub Desktop.
Sweep logs for HEAD --accuracy --backend aot_eager --training --explain (TORCHDYNAMO_DYNAMIC_SHAPES=1) - fd66568b81615212727e18052b52b845293efc24 Sat Dec 17 15:52:17 UTC 2022
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cuda train BERT_pytorch PASS | |
Dynamo produced 4 graph(s) covering 574 ops | |
cuda train Background_Matting PASS | |
Dynamo produced 2 graph(s) covering 366 ops | |
WARNING:root:DALLE2_pytorch failed to load | |
Eager model failed to run | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1019, in validate_model | |
self.model_iter_fn(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 356, in forward_and_backward_pass | |
self.grad_scaler.scale(loss).backward() | |
File "/scratch/ezyang/work/a/pytorch/torch/_tensor.py", line 484, in backward | |
torch.autograd.backward( | |
File "/scratch/ezyang/work/a/pytorch/torch/autograd/__init__.py", line 197, in backward | |
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass | |
RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 2042, in run | |
) = runner.load_model(device, model_name, batch_size=batch_size) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 300, in load_model | |
self.validate_model(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1021, in validate_model | |
raise NotImplementedError("Eager model failed to run") from e | |
NotImplementedError: Eager model failed to run | |
cuda train LearningToPaint PASS | |
Dynamo produced 2 graph(s) covering 144 ops | |
cuda train Super_SloMo PASS | |
Dynamo produced 1 graph(s) covering 374 ops | |
cuda train alexnet PASS | |
Dynamo produced 2 graph(s) covering 44 ops | |
cuda train attention_is_all_you_need_pytorch PASS | |
Dynamo produced 6 graph(s) covering 615 ops | |
cuda train dcgan PASS | |
Dynamo produced 2 graph(s) covering 26 ops | |
cuda train densenet121 PASS | |
Dynamo produced 2 graph(s) covering 862 ops | |
WARNING:root:detectron2_fcos_r_50_fpn failed to load | |
FCOS train is not supported by upstream detectron2. See GH Issue: https://github.com/facebookresearch/detectron2/issues/4369. | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 2042, in run | |
) = runner.load_model(device, model_name, batch_size=batch_size) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 268, in load_model | |
benchmark = benchmark_cls( | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/util/model.py", line 19, in __call__ | |
obj = type.__call__(cls, *args, **kwargs) | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/detectron2_fcos_r_50_fpn/__init__.py", line 15, in __init__ | |
super().__init__(variant="COCO-Detection/fcos_R_50_FPN_1x.py", test=test, device=device, | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/util/framework/detectron2/model_factory.py", line 100, in __init__ | |
loader = self.setup_train(cfg, args) | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/util/framework/detectron2/model_factory.py", line 110, in setup_train | |
raise NotImplementedError("FCOS train is not supported by upstream detectron2. " \ | |
NotImplementedError: FCOS train is not supported by upstream detectron2. See GH Issue: https://github.com/facebookresearch/detectron2/issues/4369. | |
WARNING:root:detectron2_maskrcnn_r_50_c4 failed to load | |
Eager model failed to run | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1019, in validate_model | |
self.model_iter_fn(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 355, in forward_and_backward_pass | |
loss = self.compute_loss(pred) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 345, in compute_loss | |
return reduce_to_scalar_loss(pred) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in reduce_to_scalar_loss | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in <listcomp> | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 107, in reduce_to_scalar_loss | |
return sum([reduce_to_scalar_loss(value) for value in out.values()]) / len( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 107, in <listcomp> | |
return sum([reduce_to_scalar_loss(value) for value in out.values()]) / len( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 110, in reduce_to_scalar_loss | |
raise NotImplementedError("Don't know how to reduce", type(out)) | |
NotImplementedError: ("Don't know how to reduce", <class 'detectron2.structures.instances.Instances'>) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 2042, in run | |
) = runner.load_model(device, model_name, batch_size=batch_size) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 300, in load_model | |
self.validate_model(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1021, in validate_model | |
raise NotImplementedError("Eager model failed to run") from e | |
NotImplementedError: Eager model failed to run | |
cuda train dlrm PASS | |
Dynamo produced 1 graph(s) covering 40 ops | |
WARNING:root:doctr_det_predictor failed to load | |
Eager model failed to run | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1019, in validate_model | |
self.model_iter_fn(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 355, in forward_and_backward_pass | |
loss = self.compute_loss(pred) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 345, in compute_loss | |
return reduce_to_scalar_loss(pred) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 107, in reduce_to_scalar_loss | |
return sum([reduce_to_scalar_loss(value) for value in out.values()]) / len( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 107, in <listcomp> | |
return sum([reduce_to_scalar_loss(value) for value in out.values()]) / len( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in reduce_to_scalar_loss | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in <listcomp> | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 110, in reduce_to_scalar_loss | |
raise NotImplementedError("Don't know how to reduce", type(out)) | |
NotImplementedError: ("Don't know how to reduce", <class 'numpy.ndarray'>) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 2042, in run | |
) = runner.load_model(device, model_name, batch_size=batch_size) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 300, in load_model | |
self.validate_model(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1021, in validate_model | |
raise NotImplementedError("Eager model failed to run") from e | |
NotImplementedError: Eager model failed to run | |
WARNING:root:doctr_reco_predictor failed to load | |
Eager model failed to run | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1019, in validate_model | |
self.model_iter_fn(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 355, in forward_and_backward_pass | |
loss = self.compute_loss(pred) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 345, in compute_loss | |
return reduce_to_scalar_loss(pred) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 107, in reduce_to_scalar_loss | |
return sum([reduce_to_scalar_loss(value) for value in out.values()]) / len( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 107, in <listcomp> | |
return sum([reduce_to_scalar_loss(value) for value in out.values()]) / len( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in reduce_to_scalar_loss | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in <listcomp> | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in reduce_to_scalar_loss | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 97, in <listcomp> | |
return sum([reduce_to_scalar_loss(x) for x in out]) / len(out) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/testing.py", line 110, in reduce_to_scalar_loss | |
raise NotImplementedError("Don't know how to reduce", type(out)) | |
NotImplementedError: ("Don't know how to reduce", <class 'str'>) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 2042, in run | |
) = runner.load_model(device, model_name, batch_size=batch_size) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 300, in load_model | |
self.validate_model(model, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1021, in validate_model | |
raise NotImplementedError("Eager model failed to run") from e | |
NotImplementedError: Eager model failed to run | |
/scratch/ezyang/work/a/pytorch/torch/utils/tensorboard/__init__.py:4: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead. | |
if not hasattr(tensorboard, "__version__") or LooseVersion( | |
/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/gym/core.py:317: DeprecationWarning: [33mWARN: Initializing wrapper in old step API which returns one bool instead of two. It is recommended to set `new_step_api=True` to use new step API. This will be the default behaviour in future.[0m | |
deprecation( | |
cuda train drq PASS | |
Dynamo produced 4 graph(s) covering 61 ops | |
cuda train fastNLP_Bert PASS | |
Dynamo produced 6 graph(s) covering 629 ops | |
cuda train functorch_dp_cifar10 WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//2*(Mod((s2 - 1)//2**2/((s2 - 1)//2 + 1) + 2*(s2 - 1)//2/((s2 - 1)//2 + 1) + 1/((s2 - 1)//2 + 1), 1)) + Mod((s2 - 1)//2**2/((s2 - 1)//2 + 1) + 2*(s2 - 1)//2/((s2 - 1)//2 + 1) + 1/((s2 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//4*(Mod((s2 - 1)//4**2/((s2 - 1)//4 + 1) + 2*(s2 - 1)//4/((s2 - 1)//4 + 1) + 1/((s2 - 1)//4 + 1), 1)) + Mod((s2 - 1)//4**2/((s2 - 1)//4 + 1) + 2*(s2 - 1)//4/((s2 - 1)//4 + 1) + 1/((s2 - 1)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//8*(Mod((s2 - 1)//8**2/((s2 - 1)//8 + 1) + 2*(s2 - 1)//8/((s2 - 1)//8 + 1) + 1/((s2 - 1)//8 + 1), 1)) + Mod((s2 - 1)//8**2/((s2 - 1)//8 + 1) + 2*(s2 - 1)//8/((s2 - 1)//8 + 1) + 1/((s2 - 1)//8 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//16*(Mod((s2 - 1)//16**2/((s2 - 1)//16 + 1) + 2*(s2 - 1)//16/((s2 - 1)//16 + 1) + 1/((s2 - 1)//16 + 1), 1)) + Mod((s2 - 1)//16**2/((s2 - 1)//16 + 1) + 2*(s2 - 1)//16/((s2 - 1)//16 + 1) + 1/((s2 - 1)//16 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//2*(Mod((s2 - 1)//2**2/((s2 - 1)//2 + 1) + 2*(s2 - 1)//2/((s2 - 1)//2 + 1) + 1/((s2 - 1)//2 + 1), 1)) + Mod((s2 - 1)//2**2/((s2 - 1)//2 + 1) + 2*(s2 - 1)//2/((s2 - 1)//2 + 1) + 1/((s2 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//4*(Mod((s2 - 1)//4**2/((s2 - 1)//4 + 1) + 2*(s2 - 1)//4/((s2 - 1)//4 + 1) + 1/((s2 - 1)//4 + 1), 1)) + Mod((s2 - 1)//4**2/((s2 - 1)//4 + 1) + 2*(s2 - 1)//4/((s2 - 1)//4 + 1) + 1/((s2 - 1)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//8*(Mod((s2 - 1)//8**2/((s2 - 1)//8 + 1) + 2*(s2 - 1)//8/((s2 - 1)//8 + 1) + 1/((s2 - 1)//8 + 1), 1)) + Mod((s2 - 1)//8**2/((s2 - 1)//8 + 1) + 2*(s2 - 1)//8/((s2 - 1)//8 + 1) + 1/((s2 - 1)//8 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 1)//16*(Mod((s2 - 1)//16**2/((s2 - 1)//16 + 1) + 2*(s2 - 1)//16/((s2 - 1)//16 + 1) + 1/((s2 - 1)//16 + 1), 1)) + Mod((s2 - 1)//16**2/((s2 - 1)//16 + 1) + 2*(s2 - 1)//16/((s2 - 1)//16 + 1) + 1/((s2 - 1)//16 + 1), 1) - 0, s2) | |
PASS | |
Dynamo produced 2 graph(s) covering 138 ops | |
cuda train functorch_maml_omniglot PASS | |
Dynamo produced 2 graph(s) covering 28 ops | |
cuda train hf_Albert PASS | |
Dynamo produced 4 graph(s) covering 571 ops | |
cuda train hf_Bart PASS | |
Dynamo produced 39 graph(s) covering 571 ops | |
cuda train hf_Bert PASS | |
Dynamo produced 5 graph(s) covering 556 ops | |
cuda train hf_Bert_large PASS | |
Dynamo produced 5 graph(s) covering 1096 ops | |
cuda train hf_BigBird PASS | |
Dynamo produced 63 graph(s) covering 806 ops | |
cuda train hf_DistilBert PASS | |
Dynamo produced 4 graph(s) covering 217 ops | |
cuda train hf_GPT2 PASS | |
Dynamo produced 60 graph(s) covering 924 ops | |
cuda train hf_GPT2_large PASS | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train hf_Longformer [2022-12-17 16:17:15,491] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64) | |
function: '_chunk' (/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/transformers/models/longformer/modeling_longformer.py:770) | |
reasons: hidden_states.stride()[0] == hidden_states.size()[2] and hidden_states.stride()[1] == hidden_states.size()[0]*hidden_states.size()[2] and hidden_states.stride()[2] == 1 and hidden_states.storage_offset() == 0 and Eq(hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2], 512*hidden_states.size()[0]*hidden_states.size()[2]*hidden_states.size()[1]//512) and Eq(Mod(hidden_states.size()[1], hidden_states.size()[1]//512), 0) and Ne(hidden_states.size()[1]/hidden_states.size()[1]//512, 1) and Ne(hidden_states.size()[1]//512, 1) and Ne(hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512, 1) and hidden_states.size()[1]//512 >= 2 and hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512 >= hidden_states.size()[2] and hidden_states.size()[1]/hidden_states.size()[1]//512 >= 2 and hidden_states.size()[0]*hidden_states.size()[2] < hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512 and Ne(hidden_states.size()[1]//512, 0) and hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512 >= 0 and hidden_states.size()[1]//512 > 1 and Eq(hidden_states.size()[1]/hidden_states.size()[1]//512, 512) and Ne(2*hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2] - hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512, 0) and Ne(2*hidden_states.size()[1]//512 - 1, 1) and Ne((hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2, 1) and 2*hidden_states.size()[1]//512 - 1 >= 2 and (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 >= hidden_states.size()[2] and hidden_states.size()[0]*hidden_states.size()[2] < (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 and Ne((hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2, hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512) and Ne((hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2, 0) and (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 >= 0 and 2*hidden_states.size()[1]//512 - 1 > 1 and 1 < 2*hidden_states.size()[1]//512*(hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 - (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 and hidden_states.size()[1]/hidden_states.size()[1]//512 >= 0 and Ne(2*hidden_states.size()[1]//512 - 1, -1) and 2*hidden_states.size()[1]//512 - 1 >= 0 and hidden_states.size()[0] != 0 and hidden_states.size()[0] != 1 and hidden_states.size()[1] != 0 and hidden_states.size()[1] != 1 and hidden_states.size()[2] != 0 and hidden_states.size()[2] != 1 | |
to diagnose recompilation issues, see https://github.com/pytorch/torchdynamo/blob/main/TROUBLESHOOTING.md. | |
PASS | |
Dynamo produced 124 graph(s) covering 1682 ops | |
cuda train hf_Reformer PASS | |
Dynamo produced 60 graph(s) covering 518 ops | |
cuda train hf_T5 WARNING:common:fp64 golden ref were not generated for hf_T5. Setting accuracy check to cosine | |
PASS | |
Dynamo produced 1 graph(s) covering 881 ops | |
cuda train hf_T5_base WARNING:common:fp64 golden ref were not generated for hf_T5_base. Setting accuracy check to cosine | |
PASS | |
Dynamo produced 1 graph(s) covering 1643 ops | |
cuda train hf_T5_large PASS | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train lennard_jones PASS | |
Dynamo produced 1 graph(s) covering 9 ops | |
cuda train maml_omniglot PASS | |
Dynamo produced 2 graph(s) covering 28 ops | |
cuda train mnasnet1_0 PASS | |
Dynamo produced 1 graph(s) covering 152 ops | |
cuda train mobilenet_v2 PASS | |
Dynamo produced 1 graph(s) covering 153 ops | |
cuda train mobilenet_v2_quantized_qat WARNING:common:fp64 golden ref were not generated for mobilenet_v2_quantized_qat. Setting accuracy check to cosine | |
ERROR:common:output with shape [1] doesn't match the broadcast shape [32] | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 351, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 352, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 354, in <graph break in forward_and_backward_pass> | |
pred = mod(*cloned_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/fx/graph_module.py", line 660, in call_wrapped | |
return self._wrapped_call(self, *args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/fx/graph_module.py", line 279, in __call__ | |
raise e | |
File "/scratch/ezyang/work/a/pytorch/torch/fx/graph_module.py", line 269, in __call__ | |
return super(self.cls, obj).__call__(*args, **kwargs) # type: ignore[misc] | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "<eval_with_key>.8", line 4, in forward | |
def forward(self, x : torch.Tensor) -> torch.Tensor: | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 2476, in forward | |
return compiled_fn(full_args) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 1024, in g | |
return f(*args) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 2045, in debug_compiled_function | |
return compiled_function(*args) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 1962, in compiled_function | |
original_inpt.copy_(updated_inpt) | |
RuntimeError: output with shape [1] doesn't match the broadcast shape [32] | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 1 graph(s) covering 203 ops | |
cuda train mobilenet_v3_large PASS | |
Dynamo produced 1 graph(s) covering 187 ops | |
cuda train moco [2022-12-17 16:24:06,568] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64) | |
function: '<graph break in _momentum_update_key_encoder>' (/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/moco/moco/builder.py:50) | |
reasons: ___tuple_iterator_len(___stack0) == 160 | |
to diagnose recompilation issues, see https://github.com/pytorch/torchdynamo/blob/main/TROUBLESHOOTING.md. | |
ERROR:common: | |
from user code: | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/moco/moco/builder.py", line 172, in concat_all_gather | |
torch.distributed.all_gather(tensors_gather, tensor, async_op=False) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_subclasses/fake_tensor.py", line 915, in __torch_dispatch__ | |
r = func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_ops.py", line 284, in __call__ | |
return self._op(*args, **kwargs or {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_ops.py", line 377, in _get_dispatch | |
final_key = resolve_key(self, key) | |
File "/scratch/ezyang/work/a/pytorch/torch/_ops.py", line 106, in resolve_key | |
raise NotImplementedError(f"could not find kernel for {op} at dispatch key {k}") | |
NotImplementedError: could not find kernel for c10d.allgather_.default at dispatch key DispatchKey.Meta | |
During handling of the above exception, another exception occurred: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1055, in run_node | |
return node.target(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/distributed/distributed_c10d.py", line 1429, in wrapper | |
return func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/distributed/distributed_c10d.py", line 2424, in all_gather | |
work = default_pg.allgather([tensor_list], [tensor]) | |
File "/scratch/ezyang/work/a/pytorch/torch/_subclasses/fake_tensor.py", line 920, in __torch_dispatch__ | |
return run_fallback_kernel(self, func, args, kwargs, not_implemented_error) | |
File "/scratch/ezyang/work/a/pytorch/torch/_subclasses/fake_tensor.py", line 1099, in run_fallback_kernel | |
args = tree_map(to_real_tensor, args) | |
File "/scratch/ezyang/work/a/pytorch/torch/utils/_pytree.py", line 195, in tree_map | |
return tree_unflatten([fn(i) for i in flat_args], spec) | |
File "/scratch/ezyang/work/a/pytorch/torch/utils/_pytree.py", line 195, in <listcomp> | |
return tree_unflatten([fn(i) for i in flat_args], spec) | |
File "/scratch/ezyang/work/a/pytorch/torch/_subclasses/fake_tensor.py", line 1092, in to_real_tensor | |
out = torch.zeros_like(e, device=e.fake_device) | |
RuntimeError: Cannot call strides() on tensor with symbolic sizes/strides | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1014, in get_fake_value | |
return wrap_fake_exception( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 704, in wrap_fake_exception | |
return fn() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1015, in <lambda> | |
lambda: run_node(tx.output, node, args, kwargs, nnmodule) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1064, in run_node | |
raise RuntimeError( | |
RuntimeError: Failed running call_function <function all_gather at 0x7f2ee2449dc0>(*([FakeTensor(FakeTensor(..., device='meta', size=(s0, s1, s2, s2)), cuda:0)], FakeTensor(FakeTensor(..., device='meta', size=(s0, s1, s2, s2)), cuda:0)), **{'async_op': False}): | |
Cannot call strides() on tensor with symbolic sizes/strides | |
(scroll up for backtrace) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 351, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 352, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 354, in <graph break in forward_and_backward_pass> | |
pred = mod(*cloned_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/parallel/distributed.py", line 1098, in forward | |
output = self._run_ddp_forward(*inputs, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/parallel/distributed.py", line 1051, in _run_ddp_forward | |
return module_to_run(*inputs[0], **kwargs[0]) # type: ignore[index] | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/moco/moco/builder.py", line 130, in forward | |
self._momentum_update_key_encoder() # update the key encoder | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/moco/moco/builder.py", line 133, in <graph break in forward> | |
im_k, idx_unshuffle = self._batch_shuffle_ddp(im_k) | |
File "/scratch/ezyang/work/a/pytorch/torch/autograd/grad_mode.py", line 34, in decorate_context | |
return func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/moco/moco/builder.py", line 76, in _batch_shuffle_ddp | |
x_gather = concat_all_gather(x) | |
File "/scratch/ezyang/work/a/pytorch/torch/autograd/grad_mode.py", line 34, in decorate_context | |
return func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 330, in catch_errors | |
return hijacked_callback(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 480, in _convert_frame | |
result = inner_convert(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 103, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 90, in time_wrapper | |
r = func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 339, in _convert_frame_assert | |
return _compile( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 400, in _compile | |
out_code = transform_code_object(code, transform) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/bytecode_transformation.py", line 341, in transform_code_object | |
transformations(instructions, code_options) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 387, in transform | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1684, in run | |
super().run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1015, in CALL_FUNCTION_KW | |
self.call_function(fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/torch.py", line 468, in call_function | |
tensor_variable = wrap_fx_proxy( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 733, in wrap_fx_proxy | |
return wrap_fx_proxy_cls( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 773, in wrap_fx_proxy_cls | |
example_value = get_fake_value(proxy.node, tx) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1034, in get_fake_value | |
raise TorchRuntimeError() from e | |
torch._dynamo.exc.TorchRuntimeError: | |
from user code: | |
File "/scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/moco/moco/builder.py", line 172, in concat_all_gather | |
torch.distributed.all_gather(tensors_gather, tensor, async_op=False) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 68 graph(s) covering 507 ops | |
cuda train nvidia_deeprecommender PASS | |
Dynamo produced 1 graph(s) covering 13 ops | |
cuda train phlippe_densenet PASS | |
Dynamo produced 1 graph(s) covering 186 ops | |
cuda train phlippe_resnet PASS | |
Dynamo produced 1 graph(s) covering 71 ops | |
--dataroot /scratch/ezyang/work/a/torchbenchmark/torchbenchmark/data/.data/pytorch_CycleGAN_and_pix2pix_inputs/datasets/horse2zebra --name horse2zebra --model cycle_gan --display_id 0 --n_epochs 3 --n_epochs_decay 3 --gpu_ids 0 --checkpoints_dir /scratch/ezyang/work/a/torchbenchmark/torchbenchmark/models/pytorch_CycleGAN_and_pix2pix/.data/checkpoints | |
cuda train pytorch_CycleGAN_and_pix2pix PASS | |
Dynamo produced 1 graph(s) covering 91 ops | |
cuda train pytorch_stargan PASS | |
Dynamo produced 1 graph(s) covering 60 ops | |
cuda train pytorch_struct PASS | |
Dynamo produced 1 graph(s) covering 47 ops | |
cuda train pytorch_unet PASS | |
Dynamo produced 2 graph(s) covering 270 ops | |
cuda train resnet152 PASS | |
Dynamo produced 1 graph(s) covering 515 ops | |
cuda train resnet18 PASS | |
Dynamo produced 1 graph(s) covering 69 ops | |
cuda train resnet50 PASS | |
Dynamo produced 1 graph(s) covering 175 ops | |
cuda train resnet50_quantized_qat WARNING:common:fp64 golden ref were not generated for resnet50_quantized_qat. Setting accuracy check to cosine | |
ERROR:common:output with shape [1] doesn't match the broadcast shape [64] | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 351, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 352, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 354, in <graph break in forward_and_backward_pass> | |
pred = mod(*cloned_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/fx/graph_module.py", line 660, in call_wrapped | |
return self._wrapped_call(self, *args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/fx/graph_module.py", line 279, in __call__ | |
raise e | |
File "/scratch/ezyang/work/a/pytorch/torch/fx/graph_module.py", line 269, in __call__ | |
return super(self.cls, obj).__call__(*args, **kwargs) # type: ignore[misc] | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "<eval_with_key>.8", line 4, in forward | |
def forward(self, x : torch.Tensor) -> torch.Tensor: | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 2476, in forward | |
return compiled_fn(full_args) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 1024, in g | |
return f(*args) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 2045, in debug_compiled_function | |
return compiled_function(*args) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 1962, in compiled_function | |
original_inpt.copy_(updated_inpt) | |
RuntimeError: output with shape [1] doesn't match the broadcast shape [64] | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 1 graph(s) covering 163 ops | |
cuda train resnext50_32x4d PASS | |
Dynamo produced 1 graph(s) covering 175 ops | |
cuda train shufflenet_v2_x1_0 PASS | |
Dynamo produced 1 graph(s) covering 367 ops | |
/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/gym/core.py:317: DeprecationWarning: [33mWARN: Initializing wrapper in old step API which returns one bool instead of two. It is recommended to set `new_step_api=True` to use new step API. This will be the default behaviour in future.[0m | |
deprecation( | |
/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/gym/wrappers/step_api_compatibility.py:39: DeprecationWarning: [33mWARN: Initializing environment in old step API which returns one bool instead of two. It is recommended to set `new_step_api=True` to use new step API. This will be the default behaviour in future.[0m | |
deprecation( | |
cuda train soft_actor_critic PASS | |
Dynamo produced 3 graph(s) covering 20 ops | |
cuda train speech_transformer [2022-12-17 16:31:25,292] torch._dynamo.variables.builtin: [WARNING] incorrect arg count <bound method BuiltinVariable._call_min_max of BuiltinVariable(max)> missing a required argument: 'b' and no constant handler | |
PASS | |
Dynamo produced 15 graph(s) covering 849 ops | |
cuda train squeezenet1_1 PASS | |
Dynamo produced 1 graph(s) covering 66 ops | |
cuda train tacotron2 ERROR:common:one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [4, 80, 724]], which is output 0 of AsStridedBackward0, is at version 2; expected version 0 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradient, with torch.autograd.set_detect_anomaly(True). | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 351, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 352, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 356, in <graph break in forward_and_backward_pass> | |
self.grad_scaler.scale(loss).backward() | |
File "/scratch/ezyang/work/a/pytorch/torch/_tensor.py", line 484, in backward | |
torch.autograd.backward( | |
File "/scratch/ezyang/work/a/pytorch/torch/autograd/__init__.py", line 197, in backward | |
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass | |
File "/scratch/ezyang/work/a/pytorch/torch/autograd/function.py", line 273, in apply | |
return user_fn(self, *args) | |
File "/scratch/ezyang/work/a/pytorch/torch/_functorch/aot_autograd.py", line 1871, in backward | |
list(ctx.symints) + list(ctx.saved_tensors) + list(contiguous_args) | |
RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [4, 80, 724]], which is output 0 of AsStridedBackward0, is at version 2; expected version 0 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradient, with torch.autograd.set_detect_anomaly(True). | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 11 graph(s) covering 28361 ops | |
cuda train timm_efficientdet ERROR:common: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/effdet/efficientdet.py", line 211, in forward | |
input_node = resample(input_node) | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/effdet/efficientdet.py", line 134, in forward | |
return F.interpolate( | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1055, in run_node | |
return node.target(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/functional.py", line 3924, in interpolate | |
return torch._C._nn.upsample_nearest2d(input, output_size, scale_factors) | |
RuntimeError: Cannot call sizes() on tensor with symbolic sizes/strides | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1014, in get_fake_value | |
return wrap_fake_exception( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 704, in wrap_fake_exception | |
return fn() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1015, in <lambda> | |
lambda: run_node(tx.output, node, args, kwargs, nnmodule) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1064, in run_node | |
raise RuntimeError( | |
RuntimeError: Failed running call_function <function interpolate at 0x7f9351c53160>(*(FakeTensor(FakeTensor(..., device='meta', | |
size=(s0, 88, ceiling(ceiling(ceiling(ceiling(ceiling(ceiling(ceiling(s2/2)/2)/2)/2)/2)/2)/2), ceiling(ceiling(ceiling(ceiling(ceiling(ceiling(ceiling(s2/2)/2)/2)/2)/2)/2)/2)), | |
grad_fn=<MaxPool2DWithIndicesBackward0>), cuda:0), (10, 10), None, 'nearest', None), **{'recompute_scale_factor': False}): | |
Cannot call sizes() on tensor with symbolic sizes/strides | |
(scroll up for backtrace) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 351, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 352, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 354, in <graph break in forward_and_backward_pass> | |
pred = mod(*cloned_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 333, in catch_errors | |
return callback(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 480, in _convert_frame | |
result = inner_convert(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 103, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 90, in time_wrapper | |
r = func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 339, in _convert_frame_assert | |
return _compile( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 400, in _compile | |
out_code = transform_code_object(code, transform) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/bytecode_transformation.py", line 341, in transform_code_object | |
transformations(instructions, code_options) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 387, in transform | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1684, in run | |
super().run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 182, in call_function | |
tx.call_function( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1015, in CALL_FUNCTION_KW | |
self.call_function(fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/torch.py", line 468, in call_function | |
tensor_variable = wrap_fx_proxy( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 733, in wrap_fx_proxy | |
return wrap_fx_proxy_cls( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 773, in wrap_fx_proxy_cls | |
example_value = get_fake_value(proxy.node, tx) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1034, in get_fake_value | |
raise TorchRuntimeError() from e | |
torch._dynamo.exc.TorchRuntimeError: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/effdet/efficientdet.py", line 211, in forward | |
input_node = resample(input_node) | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/effdet/efficientdet.py", line 134, in forward | |
return F.interpolate( | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train timm_efficientnet PASS | |
Dynamo produced 1 graph(s) covering 313 ops | |
cuda train timm_regnet PASS | |
Dynamo produced 1 graph(s) covering 458 ops | |
cuda train timm_resnest PASS | |
Dynamo produced 1 graph(s) covering 180 ops | |
cuda train timm_vision_transformer PASS | |
Dynamo produced 1 graph(s) covering 441 ops | |
cuda train timm_vision_transformer_large PASS | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train timm_vovnet PASS | |
Dynamo produced 1 graph(s) covering 169 ops | |
cuda train tts_angular [2022-12-17 16:53:45,348] torch._dynamo.optimizations.training: [WARNING] Unable to use Aot Autograd because of presence of LSTM | |
[2022-12-17 16:53:45,430] torch._dynamo.optimizations.training: [WARNING] Unable to use Aot Autograd because of presence of LSTM | |
[2022-12-17 16:53:45,506] torch._dynamo.optimizations.training: [WARNING] Unable to use Aot Autograd because of presence of LSTM | |
PASS | |
Dynamo produced 4 graph(s) covering 11 ops | |
cuda train vgg16 PASS | |
Dynamo produced 1 graph(s) covering 40 ops | |
cuda train vision_maskrcnn Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/torchbench.py", line 368, in <module> | |
main(TorchBenchmarkRunner(), original_dir) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1697, in main | |
return maybe_fresh_cache(run, args.cold_start_latency and args.only)( | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 863, in inner | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 2076, in run | |
runner.run_one_model( | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1360, in run_one_model | |
status = self.check_accuracy( | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1166, in check_accuracy | |
if not same( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 739, in same | |
return len(ref) == len(res) and all( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 740, in <genexpr> | |
same(ai, bi, fp64_refi, cos_similarity, tol, equal_nan, exact_dtype) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 739, in same | |
return len(ref) == len(res) and all( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 740, in <genexpr> | |
same(ai, bi, fp64_refi, cos_similarity, tol, equal_nan, exact_dtype) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 750, in same | |
same( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 807, in same | |
ref_error = rmse(fp64_ref, ref).item() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 722, in rmse | |
return torch.sqrt(torch.mean(torch.square(ref - res))) | |
RuntimeError: The size of tensor a (38) must match the size of tensor b (39) at non-singleton dimension 0 | |
ERROR | |
cuda train yolov3 [2022-12-17 16:55:28,573] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64) | |
function: 'forward' (/scratch/ezyang/work/a/pytorch/torch/nn/modules/container.py:202) | |
reasons: ___check_obj_id(self, 140578271510384) | |
to diagnose recompilation issues, see https://github.com/pytorch/torchdynamo/blob/main/TROUBLESHOOTING.md. | |
PASS | |
Dynamo produced 93 graph(s) covering 349 ops | |
cuda train AlbertForMaskedLM PASS | |
Dynamo produced 4 graph(s) covering 574 ops | |
cuda train AlbertForQuestionAnswering PASS | |
Dynamo produced 4 graph(s) covering 577 ops | |
cuda train AllenaiLongformerBase [2022-12-17 16:57:48,426] torch._dynamo.convert_frame: [WARNING] torch._dynamo hit config.cache_size_limit (64) | |
function: '_chunk' (/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/transformers/models/longformer/modeling_longformer.py:770) | |
reasons: hidden_states.stride()[0] == hidden_states.size()[2] and hidden_states.stride()[1] == hidden_states.size()[0]*hidden_states.size()[2] and hidden_states.stride()[2] == 1 and hidden_states.storage_offset() == 0 and Eq(hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2], 512*hidden_states.size()[0]*hidden_states.size()[2]*hidden_states.size()[1]//512) and Eq(Mod(hidden_states.size()[1], hidden_states.size()[1]//512), 0) and Ne(hidden_states.size()[1]/hidden_states.size()[1]//512, 1) and Ne(hidden_states.size()[1]//512, 1) and Ne(hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512, 1) and hidden_states.size()[1]//512 >= 2 and hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512 >= hidden_states.size()[2] and hidden_states.size()[1]/hidden_states.size()[1]//512 >= 2 and hidden_states.size()[0]*hidden_states.size()[2] < hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512 and Ne(hidden_states.size()[1]//512, 0) and hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512 >= 0 and hidden_states.size()[1]//512 > 1 and Eq(hidden_states.size()[1]/hidden_states.size()[1]//512, 512) and Ne(2*hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2] - hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512, 0) and Ne(2*hidden_states.size()[1]//512 - 1, 1) and Ne((hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2, 1) and 2*hidden_states.size()[1]//512 - 1 >= 2 and (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 >= hidden_states.size()[2] and hidden_states.size()[0]*hidden_states.size()[2] < (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 and Ne((hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2, hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512) and Ne((hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2, 0) and (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 >= 0 and 2*hidden_states.size()[1]//512 - 1 > 1 and 1 < 2*hidden_states.size()[1]//512*(hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 - (hidden_states.size()[1]*hidden_states.size()[0]*hidden_states.size()[2]/hidden_states.size()[1]//512)//2 and hidden_states.size()[1]/hidden_states.size()[1]//512 >= 0 and Ne(2*hidden_states.size()[1]//512 - 1, -1) and 2*hidden_states.size()[1]//512 - 1 >= 0 and hidden_states.size()[0] != 0 and hidden_states.size()[0] != 1 and hidden_states.size()[1] != 0 and hidden_states.size()[1] != 1 and hidden_states.size()[2] != 0 and hidden_states.size()[2] != 1 | |
to diagnose recompilation issues, see https://github.com/pytorch/torchdynamo/blob/main/TROUBLESHOOTING.md. | |
PASS | |
Dynamo produced 124 graph(s) covering 1685 ops | |
cuda train BartForCausalLM PASS | |
Dynamo produced 26 graph(s) covering 412 ops | |
cuda train BartForConditionalGeneration PASS | |
Dynamo produced 76 graph(s) covering 1131 ops | |
cuda train BertForMaskedLM PASS | |
Dynamo produced 5 graph(s) covering 559 ops | |
cuda train BertForQuestionAnswering PASS | |
Dynamo produced 5 graph(s) covering 569 ops | |
cuda train BlenderbotForCausalLM PASS | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train BlenderbotSmallForCausalLM PASS | |
Dynamo produced 18 graph(s) covering 279 ops | |
cuda train BlenderbotSmallForConditionalGeneration PASS | |
Dynamo produced 51 graph(s) covering 753 ops | |
cuda train CamemBert PASS | |
Dynamo produced 5 graph(s) covering 572 ops | |
cuda train DebertaForMaskedLM PASS | |
Dynamo produced 77 graph(s) covering 1027 ops | |
cuda train DebertaForQuestionAnswering PASS | |
Dynamo produced 77 graph(s) covering 1037 ops | |
cuda train DebertaV2ForMaskedLM PASS | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train DebertaV2ForQuestionAnswering PASS | |
Dynamo produced 172 graph(s) covering 2189 ops | |
WARNING:__main__:Sequence Length not defined for DistilBertForMaskedLM. Choosing 128 arbitrarily | |
cuda train DistilBertForMaskedLM PASS | |
Dynamo produced 4 graph(s) covering 221 ops | |
WARNING:__main__:Sequence Length not defined for DistilBertForQuestionAnswering. Choosing 128 arbitrarily | |
cuda train DistilBertForQuestionAnswering PASS | |
Dynamo produced 4 graph(s) covering 231 ops | |
cuda train DistillGPT2 PASS | |
Dynamo produced 30 graph(s) covering 462 ops | |
If you want to use `ElectraForCausalLM` as a standalone, add `is_decoder=True.` | |
cuda train ElectraForCausalLM PASS | |
Dynamo produced 5 graph(s) covering 562 ops | |
cuda train ElectraForQuestionAnswering PASS | |
Dynamo produced 5 graph(s) covering 568 ops | |
cuda train GPT2ForSequenceClassification PASS | |
Dynamo produced 60 graph(s) covering 924 ops | |
cuda train GoogleFnet PASS | |
Dynamo produced 28 graph(s) covering 203 ops | |
cuda train LayoutLMForMaskedLM PASS | |
Dynamo produced 4 graph(s) covering 564 ops | |
cuda train LayoutLMForSequenceClassification PASS | |
Dynamo produced 5 graph(s) covering 562 ops | |
WARNING:__main__:Sequence Length not defined for M2M100ForConditionalGeneration. Choosing 128 arbitrarily | |
cuda train M2M100ForConditionalGeneration PASS | |
Dynamo produced 101 graph(s) covering 1174 ops | |
cuda train MBartForCausalLM PASS | |
Dynamo produced 38 graph(s) covering 412 ops | |
cuda train MBartForConditionalGeneration PASS | |
Dynamo produced 100 graph(s) covering 1136 ops | |
WARNING:__main__:Sequence Length not defined for MT5ForConditionalGeneration. Choosing 128 arbitrarily | |
cuda train MT5ForConditionalGeneration WARNING:common:fp64 golden ref were not generated for MT5ForConditionalGeneration. Setting accuracy check to cosine | |
PASS | |
Dynamo produced 1 graph(s) covering 1282 ops | |
If you want to use `MegatronBertForCausalLM` as a standalone, add `is_decoder=True.` | |
cuda train MegatronBertForCausalLM PASS | |
Dynamo produced 4 graph(s) covering 1105 ops | |
cuda train MegatronBertForQuestionAnswering PASS | |
Dynamo produced 4 graph(s) covering 1111 ops | |
cuda train MobileBertForMaskedLM PASS | |
Dynamo produced 5 graph(s) covering 1822 ops | |
cuda train MobileBertForQuestionAnswering PASS | |
Dynamo produced 5 graph(s) covering 1829 ops | |
cuda train OPTForCausalLM PASS | |
Dynamo produced 38 graph(s) covering 464 ops | |
cuda train PLBartForCausalLM PASS | |
Dynamo produced 14 graph(s) covering 214 ops | |
cuda train PLBartForConditionalGeneration PASS | |
Dynamo produced 40 graph(s) covering 584 ops | |
WARNING:__main__:Sequence Length not defined for PegasusForCausalLM. Choosing 128 arbitrarily | |
cuda train PegasusForCausalLM PASS | |
Dynamo produced 39 graph(s) covering 423 ops | |
WARNING:__main__:Sequence Length not defined for PegasusForConditionalGeneration. Choosing 128 arbitrarily | |
cuda train PegasusForConditionalGeneration PASS | |
Dynamo produced 101 graph(s) covering 1140 ops | |
If you want to use `RobertaLMHeadModel` as a standalone, add `is_decoder=True.` | |
cuda train RobertaForCausalLM PASS | |
Dynamo produced 5 graph(s) covering 576 ops | |
cuda train RobertaForQuestionAnswering PASS | |
Dynamo produced 5 graph(s) covering 582 ops | |
WARNING:__main__:Sequence Length not defined for Speech2Text2ForCausalLM. Choosing 128 arbitrarily | |
cuda train Speech2Text2ForCausalLM PASS | |
Dynamo produced 15 graph(s) covering 242 ops | |
cuda train T5ForConditionalGeneration WARNING:common:fp64 golden ref were not generated for T5ForConditionalGeneration. Setting accuracy check to cosine | |
PASS | |
Dynamo produced 1 graph(s) covering 885 ops | |
cuda train T5Small WARNING:common:fp64 golden ref were not generated for T5Small. Setting accuracy check to cosine | |
PASS | |
Dynamo produced 1 graph(s) covering 885 ops | |
cuda train TrOCRForCausalLM PASS | |
Dynamo produced 26 graph(s) covering 424 ops | |
WARNING:__main__:Sequence Length not defined for XGLMForCausalLM. Choosing 128 arbitrarily | |
cuda train XGLMForCausalLM PASS | |
Dynamo produced 75 graph(s) covering 843 ops | |
cuda train XLNetLMHeadModel PASS | |
Dynamo produced 4 graph(s) covering 1014 ops | |
cuda train YituTechConvBert PASS | |
Dynamo produced 5 graph(s) covering 834 ops | |
cuda train adv_inception_v3 PASS | |
Dynamo produced 3 graph(s) covering 628 ops | |
cuda train beit_base_patch16_224 WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s4**2, 197) - 0, s4) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s4**2)//197 - 197, s4) | |
PASS | |
Dynamo produced 2 graph(s) covering 515 ops | |
cuda train botnet26t_256 PASS | |
Dynamo produced 35 graph(s) covering 414 ops | |
cuda train cait_m36_384 PASS | |
Dynamo produced 2 graph(s) covering 1546 ops | |
cuda train coat_lite_mini PASS | |
Dynamo produced 2 graph(s) covering 651 ops | |
cuda train convit_base WARNING:common:fp64 golden ref were not generated for convit_base. Setting accuracy check to cosine | |
PASS | |
Dynamo produced 58 graph(s) covering 1132 ops | |
cuda train convmixer_768_32 PASS | |
Dynamo produced 2 graph(s) covering 232 ops | |
cuda train convnext_base WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 4)//4*(Mod((s2 - 4)//4**2/((s2 - 4)//4 + 1) + 2*(s2 - 4)//4/((s2 - 4)//4 + 1) + 1/((s2 - 4)//4 + 1), 1)) + Mod((s2 - 4)//4**2/((s2 - 4)//4 + 1) + 2*(s2 - 4)//4/((s2 - 4)//4 + 1) + 1/((s2 - 4)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 4)//4 - 1)//2**2 + 2*((s2 - 4)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((s2 - 4)//4 - 1)//2*(Mod(((s2 - 4)//4 - 1)//2**2/(((s2 - 4)//4 - 1)//2 + 1) + 2*((s2 - 4)//4 - 1)//2/(((s2 - 4)//4 - 1)//2 + 1) + 1/(((s2 - 4)//4 - 1)//2 + 1), 1)) + Mod(((s2 - 4)//4 - 1)//2**2/(((s2 - 4)//4 - 1)//2 + 1) + 2*((s2 - 4)//4 - 1)//2/(((s2 - 4)//4 - 1)//2 + 1) + 1/(((s2 - 4)//4 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((((s2 - 4)//4 - 1)//2 - 1)//2**2 + 2*(((s2 - 4)//4 - 1)//2 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 4)//4 - 1)//2 - 1)//2*(Mod((((s2 - 4)//4 - 1)//2 - 1)//2**2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 2*(((s2 - 4)//4 - 1)//2 - 1)//2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 1/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1), 1)) + Mod((((s2 - 4)//4 - 1)//2 - 1)//2**2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 2*(((s2 - 4)//4 - 1)//2 - 1)//2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 1/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2**2 + 2*((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2*(Mod(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2**2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1), 1)) + Mod(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2**2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 4)//4*(Mod((s2 - 4)//4**2/((s2 - 4)//4 + 1) + 2*(s2 - 4)//4/((s2 - 4)//4 + 1) + 1/((s2 - 4)//4 + 1), 1)) + Mod((s2 - 4)//4**2/((s2 - 4)//4 + 1) + 2*(s2 - 4)//4/((s2 - 4)//4 + 1) + 1/((s2 - 4)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 4)//4 - 1)//2**2 + 2*((s2 - 4)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((s2 - 4)//4 - 1)//2*(Mod(((s2 - 4)//4 - 1)//2**2/(((s2 - 4)//4 - 1)//2 + 1) + 2*((s2 - 4)//4 - 1)//2/(((s2 - 4)//4 - 1)//2 + 1) + 1/(((s2 - 4)//4 - 1)//2 + 1), 1)) + Mod(((s2 - 4)//4 - 1)//2**2/(((s2 - 4)//4 - 1)//2 + 1) + 2*((s2 - 4)//4 - 1)//2/(((s2 - 4)//4 - 1)//2 + 1) + 1/(((s2 - 4)//4 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((((s2 - 4)//4 - 1)//2 - 1)//2**2 + 2*(((s2 - 4)//4 - 1)//2 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 4)//4 - 1)//2 - 1)//2*(Mod((((s2 - 4)//4 - 1)//2 - 1)//2**2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 2*(((s2 - 4)//4 - 1)//2 - 1)//2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 1/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1), 1)) + Mod((((s2 - 4)//4 - 1)//2 - 1)//2**2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 2*(((s2 - 4)//4 - 1)//2 - 1)//2/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1) + 1/((((s2 - 4)//4 - 1)//2 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2**2 + 2*((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2*(Mod(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2**2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1), 1)) + Mod(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2**2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 4)//4 - 1)//2 - 1)//2 - 1)//2 + 1), 1) - 0, s2) | |
PASS | |
Dynamo produced 3 graph(s) covering 1048 ops | |
cuda train crossvit_9_240 ERROR:common: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/crossvit.py", line 394, in forward_features | |
x_ = scale_image(x_, ss, self.crop_scale) | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/crossvit.py", line 281, in scale_image | |
x = torch.nn.functional.interpolate(x, size=ss, mode='bicubic', align_corners=False) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1055, in run_node | |
return node.target(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/functional.py", line 3960, in interpolate | |
return torch._C._nn.upsample_bicubic2d(input, output_size, align_corners, scale_factors) | |
RuntimeError: Cannot call sizes() on tensor with symbolic sizes/strides | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1014, in get_fake_value | |
return wrap_fake_exception( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 704, in wrap_fake_exception | |
return fn() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1015, in <lambda> | |
lambda: run_node(tx.output, node, args, kwargs, nnmodule) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1064, in run_node | |
raise RuntimeError( | |
RuntimeError: Failed running call_function <function interpolate at 0x7f8e72ec8160>(*(FakeTensor(FakeTensor(..., device='meta', size=(s0, 3, 240, 240)), cuda:0),), **{'size': (224, 224), 'mode': 'bicubic', 'align_corners': False}): | |
Cannot call sizes() on tensor with symbolic sizes/strides | |
(scroll up for backtrace) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 315, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 316, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 318, in <graph break in forward_and_backward_pass> | |
pred = mod(*cloned_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 333, in catch_errors | |
return callback(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 480, in _convert_frame | |
result = inner_convert(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 103, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 90, in time_wrapper | |
r = func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 339, in _convert_frame_assert | |
return _compile( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 400, in _compile | |
out_code = transform_code_object(code, transform) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/bytecode_transformation.py", line 341, in transform_code_object | |
transformations(instructions, code_options) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 387, in transform | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1684, in run | |
super().run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 244, in call_function | |
return super().call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 214, in call_function | |
return super(UserFunctionVariable, self).call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 67, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 214, in call_function | |
return super(UserFunctionVariable, self).call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 67, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1015, in CALL_FUNCTION_KW | |
self.call_function(fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/torch.py", line 468, in call_function | |
tensor_variable = wrap_fx_proxy( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 733, in wrap_fx_proxy | |
return wrap_fx_proxy_cls( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 773, in wrap_fx_proxy_cls | |
example_value = get_fake_value(proxy.node, tx) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1034, in get_fake_value | |
raise TorchRuntimeError() from e | |
torch._dynamo.exc.TorchRuntimeError: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/crossvit.py", line 394, in forward_features | |
x_ = scale_image(x_, ss, self.crop_scale) | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/crossvit.py", line 281, in scale_image | |
x = torch.nn.functional.interpolate(x, size=ss, mode='bicubic', align_corners=False) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train cspdarknet53 PASS | |
Dynamo produced 41 graph(s) covering 407 ops | |
cuda train deit_base_distilled_patch16_224 PASS | |
Dynamo produced 2 graph(s) covering 449 ops | |
cuda train dla102 PASS | |
Dynamo produced 3 graph(s) covering 832 ops | |
cuda train dm_nfnet_f0 PASS | |
Dynamo produced 3 graph(s) covering 1502 ops | |
cuda train dpn107 PASS | |
Dynamo produced 2 graph(s) covering 744 ops | |
cuda train eca_botnext26ts_256 PASS | |
Dynamo produced 33 graph(s) covering 430 ops | |
cuda train eca_halonext26ts WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s3**2, 64) - 0, s3) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, ((128*s0)//((s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))*((s2 - 1)//2 + 1)//(s2/8)**2*(s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))//(128*s0)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, 4) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 + 1)//(s2/8)**2)//4 - 4, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod((((s2 - 1)//2 + 1)//(s2/8)**2)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, ((128*s0)//((s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))*((s2 - 1)//2 + 1)//(s2/8)**2*(s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))//(128*s0)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, 4) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 + 1)//(s2/8)**2)//4 - 4, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod((((s2 - 1)//2 + 1)//(s2/8)**2)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((128*s0)//((s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))*((s2 - 1)//2**2 + 2*(s2 - 1)//2 + 1)//(s2**2/32), 128) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s3**2, 16) - 0, s3) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s3**2, 64) - 0, s3) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s0**3, 64) - 0, s0) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, ((128*s0)//((s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))*((s2 - 1)//2 + 1)//(s2/8)**2*(s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))//(128*s0)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, 4) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 + 1)//(s2/8)**2)//4 - 4, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod((((s2 - 1)//2 + 1)//(s2/8)**2)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, ((128*s0)//((s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))*((s2 - 1)//2 + 1)//(s2/8)**2*(s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))//(128*s0)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 + 1)//(s2/8)**2, 4) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 + 1)//(s2/8)**2)//4 - 4, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod((((s2 - 1)//2 + 1)//(s2/8)**2)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((128*s0)//((s0*(s2 - 1)//2**2 + 2*s0*(s2 - 1)//2 + s0)//(s2**2/32))*((s2 - 1)//2**2 + 2*(s2 - 1)//2 + 1)//(s2**2/32), 128) - 0, s2) | |
PASS | |
Dynamo produced 36 graph(s) covering 493 ops | |
cuda train ese_vovnet19b_dw PASS | |
Dynamo produced 2 graph(s) covering 158 ops | |
cuda train fbnetc_100 PASS | |
Dynamo produced 2 graph(s) covering 381 ops | |
cuda train fbnetv3_b PASS | |
Dynamo produced 2 graph(s) covering 608 ops | |
cuda train gernet_l PASS | |
Dynamo produced 2 graph(s) covering 407 ops | |
cuda train ghostnet_100 PASS | |
Dynamo produced 2 graph(s) covering 326 ops | |
cuda train gluon_inception_v3 PASS | |
Dynamo produced 3 graph(s) covering 628 ops | |
cuda train gluon_xception65 [2022-12-17 17:58:35,313] torch._dynamo.utils: [ERROR] RMSE (res-fp64): 0.01022, (ref-fp64): 0.00320 and shape=torch.Size([728]) | |
[2022-12-17 17:58:35,313] torch._dynamo.utils: [ERROR] Accuracy failed for key name mid.block19.rep.bn1.weight.grad | |
FAIL | |
Dynamo produced 2 graph(s) covering 354 ops | |
cuda train gmixer_24_224 PASS | |
Dynamo produced 2 graph(s) covering 640 ops | |
cuda train gmlp_s16_224 PASS | |
Dynamo produced 2 graph(s) covering 496 ops | |
cuda train hrnet_w18 ERROR:common: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/hrnet.py", line 713, in stages | |
yl = self.stage2(xl) | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/hrnet.py", line 495, in forward | |
y = y + fuse_outer[j](x[j]) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1060, in run_node | |
return nnmodule(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/upsampling.py", line 156, in forward | |
return F.interpolate(input, self.size, self.scale_factor, self.mode, self.align_corners, | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/functional.py", line 3924, in interpolate | |
return torch._C._nn.upsample_nearest2d(input, output_size, scale_factors) | |
RuntimeError: Cannot call sizes() on tensor with symbolic sizes/strides | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1014, in get_fake_value | |
return wrap_fake_exception( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 704, in wrap_fake_exception | |
return fn() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1015, in <lambda> | |
lambda: run_node(tx.output, node, args, kwargs, nnmodule) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1064, in run_node | |
raise RuntimeError( | |
RuntimeError: Failed running call_module sub1_1_2(*(FakeTensor(FakeTensor(..., device='meta', size=(s0, 18, (s2 - 1)//8 + 1, (s2 - 1)//8 + 1), | |
grad_fn=<NativeBatchNormLegitBackward0>), cuda:0),), **{}): | |
Cannot call sizes() on tensor with symbolic sizes/strides | |
(scroll up for backtrace) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 315, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 316, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 318, in <graph break in forward_and_backward_pass> | |
pred = mod(*cloned_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 333, in catch_errors | |
return callback(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 480, in _convert_frame | |
result = inner_convert(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 103, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 90, in time_wrapper | |
r = func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 339, in _convert_frame_assert | |
return _compile( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 400, in _compile | |
out_code = transform_code_object(code, transform) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/bytecode_transformation.py", line 341, in transform_code_object | |
transformations(instructions, code_options) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 387, in transform | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1684, in run | |
super().run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 244, in call_function | |
return super().call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 214, in call_function | |
return super(UserFunctionVariable, self).call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 67, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 244, in call_function | |
return super().call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 214, in call_function | |
return super(UserFunctionVariable, self).call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 67, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 182, in call_function | |
tx.call_function( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 182, in call_function | |
tx.call_function( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 201, in call_function | |
return wrap_fx_proxy( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 733, in wrap_fx_proxy | |
return wrap_fx_proxy_cls( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 773, in wrap_fx_proxy_cls | |
example_value = get_fake_value(proxy.node, tx) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1034, in get_fake_value | |
raise TorchRuntimeError() from e | |
torch._dynamo.exc.TorchRuntimeError: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/hrnet.py", line 713, in stages | |
yl = self.stage2(xl) | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/hrnet.py", line 495, in forward | |
y = y + fuse_outer[j](x[j]) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train inception_v3 PASS | |
Dynamo produced 3 graph(s) covering 628 ops | |
cuda train jx_nest_base PASS | |
Dynamo produced 2 graph(s) covering 1240 ops | |
cuda train lcnet_050 PASS | |
Dynamo produced 2 graph(s) covering 166 ops | |
cuda train levit_128 ERROR:common: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/levit.py", line 293, in forward | |
q, k, v = self.qkv(x).view( | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/levit.py", line 172, in forward | |
return self.bn(x.flatten(0, 1)).reshape_as(x) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1057, in run_node | |
return getattr(args[0], node.target)(*args[1:], **kwargs) | |
RuntimeError: Cannot call sizes() on tensor with symbolic sizes/strides | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1014, in get_fake_value | |
return wrap_fake_exception( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 704, in wrap_fake_exception | |
return fn() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1015, in <lambda> | |
lambda: run_node(tx.output, node, args, kwargs, nnmodule) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1064, in run_node | |
raise RuntimeError( | |
RuntimeError: Failed running call_method reshape_as(*(FakeTensor(FakeTensor(..., device='meta', | |
size=(s0*(s2 - 1)//16**2 + 2*s0*(s2 - 1)//16 + s0, 256), | |
grad_fn=<NativeBatchNormLegitBackward0>), cuda:0), FakeTensor(FakeTensor(..., device='meta', | |
size=(s0, (s2 - 1)//16**2 + 2*(s2 - 1)//16 + 1, 256), | |
grad_fn=<ViewBackward0>), cuda:0)), **{}): | |
Cannot call sizes() on tensor with symbolic sizes/strides | |
(scroll up for backtrace) | |
The above exception was the direct cause of the following exception: | |
Traceback (most recent call last): | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1184, in check_accuracy | |
new_result = optimized_model_iter_fn(model_copy, example_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 212, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/common.py", line 1061, in run_n_iterations | |
self.model_iter_fn(mod, inputs, collect_outputs=False) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 315, in forward_and_backward_pass | |
cloned_inputs = clone_inputs(inputs) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 316, in <graph break in forward_and_backward_pass> | |
self.optimizer_zero_grad(mod) | |
File "/scratch/ezyang/work/a/pytorch/benchmarks/dynamo/timm_models.py", line 318, in <graph break in forward_and_backward_pass> | |
pred = mod(*cloned_inputs) | |
File "/scratch/ezyang/work/a/pytorch/torch/nn/modules/module.py", line 1482, in _call_impl | |
return forward_call(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/eval_frame.py", line 333, in catch_errors | |
return callback(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 480, in _convert_frame | |
result = inner_convert(frame, cache_size, hooks) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 103, in _fn | |
return fn(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 90, in time_wrapper | |
r = func(*args, **kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 339, in _convert_frame_assert | |
return _compile( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 400, in _compile | |
out_code = transform_code_object(code, transform) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/bytecode_transformation.py", line 341, in transform_code_object | |
transformations(instructions, code_options) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/convert_frame.py", line 387, in transform | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1684, in run | |
super().run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 244, in call_function | |
return super().call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 214, in call_function | |
return super(UserFunctionVariable, self).call_function(tx, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/functions.py", line 67, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 182, in call_function | |
tx.call_function( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/nn_module.py", line 220, in call_function | |
return tx.inline_user_function_return( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 471, in inline_user_function_return | |
result = InliningInstructionTranslator.inline_call(self, fn, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1762, in inline_call | |
return cls.inline_call_(parent, func, args, kwargs) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 1817, in inline_call_ | |
tracer.run() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 538, in run | |
and self.step() | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 501, in step | |
getattr(self, inst.opname)(inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 307, in wrapper | |
return inner_fn(self, inst) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 966, in CALL_FUNCTION | |
self.call_function(fn, args, {}) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/symbolic_convert.py", line 435, in call_function | |
self.push(fn.call_function(self, args, kwargs)) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/misc.py", line 598, in call_function | |
return self.obj.call_method(tx, self.name, args, kwargs).add_options(self) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/tensor.py", line 368, in call_method | |
return wrap_fx_proxy( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 733, in wrap_fx_proxy | |
return wrap_fx_proxy_cls( | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/variables/builder.py", line 773, in wrap_fx_proxy_cls | |
example_value = get_fake_value(proxy.node, tx) | |
File "/scratch/ezyang/work/a/pytorch/torch/_dynamo/utils.py", line 1034, in get_fake_value | |
raise TorchRuntimeError() from e | |
torch._dynamo.exc.TorchRuntimeError: | |
from user code: | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/levit.py", line 293, in forward | |
q, k, v = self.qkv(x).view( | |
File "/data/home/ezyang/local/a/pytorch-env/lib/python3.9/site-packages/timm/models/levit.py", line 172, in forward | |
return self.bn(x.flatten(0, 1)).reshape_as(x) | |
Set torch._dynamo.config.verbose=True for more information | |
You can suppress this exception and fall back to eager by setting: | |
torch._dynamo.config.suppress_errors = True | |
TorchDynamo optimized model failed to run because of following error | |
FAIL | |
Dynamo produced 0 graph(s) covering 0 ops | |
cuda train mixer_b16_224 PASS | |
Dynamo produced 2 graph(s) covering 232 ops | |
cuda train mixnet_l PASS | |
Dynamo produced 2 graph(s) covering 675 ops | |
cuda train mnasnet_100 PASS | |
Dynamo produced 2 graph(s) covering 302 ops | |
cuda train mobilenetv2_100 PASS | |
Dynamo produced 2 graph(s) covering 302 ops | |
cuda train mobilenetv3_large_100 PASS | |
Dynamo produced 2 graph(s) covering 313 ops | |
cuda train mobilevit_s PASS | |
Dynamo produced 2 graph(s) covering 631 ops | |
cuda train nfnet_l0 PASS | |
Dynamo produced 2 graph(s) covering 548 ops | |
cuda train pit_b_224 PASS | |
Dynamo produced 8 graph(s) covering 494 ops | |
cuda train pnasnet5large PASS | |
Dynamo produced 3 graph(s) covering 4198 ops | |
cuda train poolformer_m36 WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//4*(Mod((s2 - 3)//4**2/((s2 - 3)//4 + 1) + 2*(s2 - 3)//4/((s2 - 3)//4 + 1) + 1/((s2 - 3)//4 + 1), 1)) + Mod((s2 - 3)//4**2/((s2 - 3)//4 + 1) + 2*(s2 - 3)//4/((s2 - 3)//4 + 1) + 1/((s2 - 3)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//8*(Mod((s2 - 3)//8**2/((s2 - 3)//8 + 1) + 2*(s2 - 3)//8/((s2 - 3)//8 + 1) + 1/((s2 - 3)//8 + 1), 1)) + Mod((s2 - 3)//8**2/((s2 - 3)//8 + 1) + 2*(s2 - 3)//8/((s2 - 3)//8 + 1) + 1/((s2 - 3)//8 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//16*(Mod((s2 - 3)//16**2/((s2 - 3)//16 + 1) + 2*(s2 - 3)//16/((s2 - 3)//16 + 1) + 1/((s2 - 3)//16 + 1), 1)) + Mod((s2 - 3)//16**2/((s2 - 3)//16 + 1) + 2*(s2 - 3)//16/((s2 - 3)//16 + 1) + 1/((s2 - 3)//16 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//32*(Mod((s2 - 3)//32**2/((s2 - 3)//32 + 1) + 2*(s2 - 3)//32/((s2 - 3)//32 + 1) + 1/((s2 - 3)//32 + 1), 1)) + Mod((s2 - 3)//32**2/((s2 - 3)//32 + 1) + 2*(s2 - 3)//32/((s2 - 3)//32 + 1) + 1/((s2 - 3)//32 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//4*(Mod((s2 - 3)//4**2/((s2 - 3)//4 + 1) + 2*(s2 - 3)//4/((s2 - 3)//4 + 1) + 1/((s2 - 3)//4 + 1), 1)) + Mod((s2 - 3)//4**2/((s2 - 3)//4 + 1) + 2*(s2 - 3)//4/((s2 - 3)//4 + 1) + 1/((s2 - 3)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//8*(Mod((s2 - 3)//8**2/((s2 - 3)//8 + 1) + 2*(s2 - 3)//8/((s2 - 3)//8 + 1) + 1/((s2 - 3)//8 + 1), 1)) + Mod((s2 - 3)//8**2/((s2 - 3)//8 + 1) + 2*(s2 - 3)//8/((s2 - 3)//8 + 1) + 1/((s2 - 3)//8 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//16*(Mod((s2 - 3)//16**2/((s2 - 3)//16 + 1) + 2*(s2 - 3)//16/((s2 - 3)//16 + 1) + 1/((s2 - 3)//16 + 1), 1)) + Mod((s2 - 3)//16**2/((s2 - 3)//16 + 1) + 2*(s2 - 3)//16/((s2 - 3)//16 + 1) + 1/((s2 - 3)//16 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s2 - 3)//32*(Mod((s2 - 3)//32**2/((s2 - 3)//32 + 1) + 2*(s2 - 3)//32/((s2 - 3)//32 + 1) + 1/((s2 - 3)//32 + 1), 1)) + Mod((s2 - 3)//32**2/((s2 - 3)//32 + 1) + 2*(s2 - 3)//32/((s2 - 3)//32 + 1) + 1/((s2 - 3)//32 + 1), 1) - 0, s2) | |
PASS | |
Dynamo produced 3 graph(s) covering 1392 ops | |
cuda train regnety_002 PASS | |
Dynamo produced 2 graph(s) covering 366 ops | |
cuda train repvgg_a2 PASS | |
Dynamo produced 2 graph(s) covering 395 ops | |
cuda train res2net101_26w_4s PASS | |
Dynamo produced 2 graph(s) covering 805 ops | |
cuda train res2net50_14w_8s PASS | |
Dynamo produced 2 graph(s) covering 701 ops | |
cuda train res2next50 PASS | |
Dynamo produced 2 graph(s) covering 397 ops | |
cuda train resmlp_12_224 PASS | |
Dynamo produced 2 graph(s) covering 208 ops | |
cuda train resnest101e PASS | |
Dynamo produced 2 graph(s) covering 1284 ops | |
cuda train rexnet_100 PASS | |
Dynamo produced 2 graph(s) covering 402 ops | |
cuda train sebotnet33ts_256 PASS | |
Dynamo produced 45 graph(s) covering 564 ops | |
cuda train selecsls42b PASS | |
Dynamo produced 2 graph(s) covering 134 ops | |
cuda train spnasnet_100 PASS | |
Dynamo produced 2 graph(s) covering 374 ops | |
cuda train swin_base_patch4_window7_224 WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s3**2, 49) - 0, s3) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s3**2)//49 - 49, s3) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s3**2, 49) - 0, s3) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((s3**2)//49 - 49, s3) | |
PASS | |
Dynamo produced 3 graph(s) covering 4024 ops | |
cuda train swsl_resnext101_32x16d PASS | |
Dynamo produced 2 graph(s) covering 413 ops | |
cuda train tf_efficientnet_b0 PASS | |
Dynamo produced 3 graph(s) covering 1086 ops | |
cuda train tf_mixnet_l PASS | |
Dynamo produced 3 graph(s) covering 2428 ops | |
cuda train tinynet_a PASS | |
Dynamo produced 2 graph(s) covering 433 ops | |
cuda train tnt_s_patch16_224 PASS | |
Dynamo produced 2 graph(s) covering 956 ops | |
cuda train twins_pcpvt_base WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1, s2//4) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(64*s0*(s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + 64*s0*(s2//4 - 8)//8 + 64*s0*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(64*s0*(Mod((s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + (s2//4 - 8)//8 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + (s2//4 - 8)//8 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(128*s0*(Mod((s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + (s2//4 - 8)//8 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(64*s0*s2//4*((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(s2//4*((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s2//4*((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1, s2//8) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(128*s0*(s2//8 - 4)//4*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4 + 128*s0*(s2//8 - 4)//4 + 128*s0*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(128*s0*(Mod((s2//8 - 4)//4*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4 + (s2//8 - 4)//4 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((s2//8 - 4)//4*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4 + (s2//8 - 4)//4 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(256*s0*(Mod((s2//8 - 4)//4*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4 + (s2//8 - 4)//4 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 4)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(128*s0*s2//8*((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(s2//8*((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1, s2//16) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(320*s0*(s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + 320*s0*(s2//16 - 2)//2 + 320*s0*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(320*s0*(Mod((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(640*s0*(Mod((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(320*s0*s2//16*((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(s2//16*((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(1024*s0*(Mod((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + 1, s2//32) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(512*s0*s2//32*((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + 1)//(s2//32), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(s2//32*((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + 1)//(s2//32), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(1024*s0*(Mod(s2//32*((s2//16 - 2)//2*(((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + (s2//16 - 2)//2 + (((s2//8 - 2)//2*(((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + (s2//8 - 2)//2 + (((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1)//(s2//8) - 2)//2 + 1)//(s2//16) - 2)//2 + 1)//(s2//32), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1, s2//4) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(64*s0*(s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + 64*s0*(s2//4 - 8)//8 + 64*s0*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(64*s0*(Mod((s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + (s2//4 - 8)//8 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + (s2//4 - 8)//8 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(128*s0*(Mod((s2//4 - 8)//8*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8 + (s2//4 - 8)//8 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 8)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(4*(Mod(64*s0*s2//4*((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(s2//4*((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(s2//4*((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod((s2//4 - 2)//2*(((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + (s2//4 - 2)//2 + (((s2 - 4)//4**2 + 2*(s2 - 4)//4 + 1)//(s2//4) - 2)//2 + 1, s2//8) - 0, s2) | |
TIMEOUT | |
cuda train visformer_small WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(1152*s0*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2 + 2*(((s2 - 1)//2 - 3)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 - 3)//4 - 1)//2*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1)) + Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(2304*s0*(Mod(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2**2 + 2*((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2*(Mod(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2**2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1), 1)) + Mod(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2**2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(1152*s0*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2 + 2*(((s2 - 1)//2 - 3)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 - 3)//4 - 1)//2*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1)) + Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(2304*s0*(Mod(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2**2 + 2*((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2*(Mod(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2**2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1), 1)) + Mod(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2**2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 2*((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1) + 1/(((((s2 - 1)//2 - 3)//4 - 1)//2 - 1)//2 + 1), 1) - 0, s2) | |
PASS | |
Dynamo produced 3 graph(s) covering 734 ops | |
cuda train vit_base_patch16_224 PASS | |
Dynamo produced 2 graph(s) covering 443 ops | |
cuda train volo_d1_224 WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 1)//2 - 3)//4**2 + 2*((s2 - 1)//2 - 3)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((s2 - 1)//2 - 3)//4*(Mod(((s2 - 1)//2 - 3)//4**2/(((s2 - 1)//2 - 3)//4 + 1) + 2*((s2 - 1)//2 - 3)//4/(((s2 - 1)//2 - 3)//4 + 1) + 1/(((s2 - 1)//2 - 3)//4 + 1), 1)) + Mod(((s2 - 1)//2 - 3)//4**2/(((s2 - 1)//2 - 3)//4 + 1) + 2*((s2 - 1)//2 - 3)//4/(((s2 - 1)//2 - 3)//4 + 1) + 1/(((s2 - 1)//2 - 3)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((s2 - 1)//2 - 3)//8*(Mod(((s2 - 1)//2 - 3)//8**2/(((s2 - 1)//2 - 3)//8 + 1) + 2*((s2 - 1)//2 - 3)//8/(((s2 - 1)//2 - 3)//8 + 1) + 1/(((s2 - 1)//2 - 3)//8 + 1), 1)) + Mod(((s2 - 1)//2 - 3)//8**2/(((s2 - 1)//2 - 3)//8 + 1) + 2*((s2 - 1)//2 - 3)//8/(((s2 - 1)//2 - 3)//8 + 1) + 1/(((s2 - 1)//2 - 3)//8 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8 + 1, ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(6*(Mod(((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2 + 2*(((s2 - 1)//2 - 3)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 - 3)//4 - 1)//2*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1)) + Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(384*s0*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2 + 2*(((s2 - 1)//2 - 3)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(6*s0*(Mod(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2, ((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8 + 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(6*(Mod(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2, ((s2 - 1)//2 - 3)//8 + 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 1)//2 - 3)//8*(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2)//(((s2 - 1)//2 - 3)//8 + 1) + (ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2)//(((s2 - 1)//2 - 3)//8 + 1), 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 1)//2 - 3)//4**2 + 2*((s2 - 1)//2 - 3)//4, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((s2 - 1)//2 - 3)//4*(Mod(((s2 - 1)//2 - 3)//4**2/(((s2 - 1)//2 - 3)//4 + 1) + 2*((s2 - 1)//2 - 3)//4/(((s2 - 1)//2 - 3)//4 + 1) + 1/(((s2 - 1)//2 - 3)//4 + 1), 1)) + Mod(((s2 - 1)//2 - 3)//4**2/(((s2 - 1)//2 - 3)//4 + 1) + 2*((s2 - 1)//2 - 3)//4/(((s2 - 1)//2 - 3)//4 + 1) + 1/(((s2 - 1)//2 - 3)//4 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(((s2 - 1)//2 - 3)//8*(Mod(((s2 - 1)//2 - 3)//8**2/(((s2 - 1)//2 - 3)//8 + 1) + 2*((s2 - 1)//2 - 3)//8/(((s2 - 1)//2 - 3)//8 + 1) + 1/(((s2 - 1)//2 - 3)//8 + 1), 1)) + Mod(((s2 - 1)//2 - 3)//8**2/(((s2 - 1)//2 - 3)//8 + 1) + 2*((s2 - 1)//2 - 3)//8/(((s2 - 1)//2 - 3)//8 + 1) + 1/(((s2 - 1)//2 - 3)//8 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8 + 1, ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(6*(Mod(((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2 + 2*(((s2 - 1)//2 - 3)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve((((s2 - 1)//2 - 3)//4 - 1)//2*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1)) + Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 2*(((s2 - 1)//2 - 3)//4 - 1)//2/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1) + 1/((((s2 - 1)//2 - 3)//4 - 1)//2 + 1), 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(384*s0*(Mod((((s2 - 1)//2 - 3)//4 - 1)//2**2 + 2*(((s2 - 1)//2 - 3)//4 - 1)//2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(6*s0*(Mod(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2, ((s2 - 1)//2 - 3)//8**2 + 2*((s2 - 1)//2 - 3)//8 + 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(6*(Mod(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2, 1)) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(Mod(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2, ((s2 - 1)//2 - 3)//8 + 1) - 0, s2) | |
WARNING:torch.fx.experimental.symbolic_shapes:RecursionError in sympy.solve(s0*(Mod(((s2 - 1)//2 - 3)//8*(ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2)//(((s2 - 1)//2 - 3)//8 + 1) + (ceiling(((s2 - 1)//2 - 3)//4/2 + 1/2)**2)//(((s2 - 1)//2 - 3)//8 + 1), 1)) - 0, s2) | |
[2022-12-17 19:09:03,297] torch._dynamo.utils: [ERROR] RMSE (res-fp64): 0.00361, (ref-fp64): 0.00018 and shape=torch.Size([64, 3, 7, 7]) | |
[2022-12-17 19:09:03,297] torch._dynamo.utils: [ERROR] Accuracy failed for key name patch_embed.conv.0.weight.grad | |
FAIL | |
Dynamo produced 3 graph(s) covering 1568 ops | |
cuda train xcit_large_24_p8_224 WARNING:common:fp64 golden ref were not generated for xcit_large_24_p8_224. Setting accuracy check to cosine | |
FAIL | |
Dynamo produced 0 graph(s) covering 0 ops |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment