RuntimeError: Inference tensors do not track version counter.

最新推荐文章于 2025-03-10 10:19:40 发布

原创

最新推荐文章于 2025-03-10 10:19:40 发布 · 1.4k 阅读

1 ·

CC 4.0 BY-SA版权

文章标签：

#pytorch #lighting #trainer.test

文章描述了在使用PyTorch2.0的torch.compile()函数编译模型时遇到的问题，涉及版本计数器丢失导致的运行时错误，解决方法是移除编译步骤。

问题：

Testing DataLoader 0:   0%|          | 0/78 [00:00<?, ?it/s]Failed to collect metadata on function, produced code may be suboptimal.  Known situations this can occur are inference mode only compilation involving resize_ or prims (!schema.hasAnyAliasInfo() INTERNAL ASSERT FAILED); if your situation looks different please file a bug to PyTorch.
Traceback (most recent call last):
  File "/home/buty/.local/lib/python3.8/site-packages/torch/_functorch/aot_autograd.py", line 1674, in aot_wrapper_dedupe
    fw_metadata, _out = run_functionalized_fw_and_collect_metadata(
  File "/home/buty/.local/lib/python3.8/site-packages/torch/_functorch/aot_autograd.py", line 606, in inner
    flat_f_outs = f(*flat_f_args)
  File "/home/buty/.local/lib/python3.8/site-packages/torch/_functorch/aot_autograd.py", line 2776, in functional_call
    out = Interpreter(mod).run(*args[params_len:], **kwargs)
  File "/home/buty/.local/lib/python3.8/site-packages/torch/fx/interpreter.py", line 136, in run
    self.env[node] = self.run_node(node)
  File "/home/buty/.local/lib/python3.8/site-packages/torch/fx/interpreter.py", line 177, in run_node
    return getattr(self, n.op)(n.target, args, kwargs)
  File "/home/buty/.local/lib/python3.8/s