Spaces:

open-llm-leaderboard

/

open_llm_leaderboard

like 11.9k

Running on CPU Upgrade

Hi!
Your model failed because it could not be loaded, did you follow all the steps in the About ?

  File "...python3.10/site-packages/transformers/models/auto/auto_factory.py", line 566, in from_pretrained
    return model_class.from_pretrained(
  File "...python3.10/site-packages/transformers/modeling_utils.py", line 3761, in from_pretrained
    dispatch_model(model, **device_map_kwargs)
  File "...python3.10/site-packages/accelerate/big_modeling.py", line 421, in dispatch_model
    model.to(device)
  File "...python3.10/site-packages/transformers/modeling_utils.py", line 2460, in to
    return super().to(*args, **kwargs)
  File "...python3.10/site-packages/torch/nn/modules/module.py", line 1145, in to
    return self._apply(convert)
  File "...python3.10/site-packages/torch/nn/modules/module.py", line 797, in _apply
    module._apply(fn)
  File "...python3.10/site-packages/torch/nn/modules/module.py", line 797, in _apply
    module._apply(fn)
  File "...python3.10/site-packages/torch/nn/modules/module.py", line 797, in _apply
    module._apply(fn)
  [Previous line repeated 2 more times]
  File "...python3.10/site-packages/torch/nn/modules/module.py", line 820, in _apply
    param_applied = fn(param)
  File "...python3.10/site-packages/torch/nn/modules/module.py", line 1143, in convert
    return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)
NotImplementedError: Cannot copy out of meta tensor; no data!

Hello again, thank you for sharing this. I am not sure why this happened but I assume my super bad internet may have something to do with it. Solved!

Edit: My other merge got a score, and I didn't do anything different. I am barely scratching the surface of LLMs and Python, so excuse my lack of knowledge if I failed to follow the steps!

It's probably not a matter of your internet connection since we load models from the hub.

Regarding the steps, we suggest to do a number of things before submitting to ensure model evaluation goes smoothly (converting the model with safetensors, making sure it can be loaded with from_pretrained) - once you followed them for your model, I can relaunch its evaluation if you want.

Thank you for your comment!

It's probably not a matter of your internet connection since we load models from the hub.

Regarding the steps, we suggest to do a number of things before submitting to ensure model evaluation goes smoothly (converting the model with safetensors, making sure it can be loaded with from_pretrained) - once you followed them for your model, I can relaunch its evaluation if you want.

Thank you for your comment!

True, what I failed to mention was that I probably disconnected while uploading a shard due to bad connection. I checked the SHA of safetensors. The last one was wrong:

I now fixed it and loaded the model successfully with the provided code:

I would appreciate if you could try processing again!

I see, that makes sense!
I relaunched it :)