You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
RuntimeError: CUDA error: out of memory
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.`
on colab it work perfectly.
any idea ?
The text was updated successfully, but these errors were encountered:
It could be something related to the CUDA version this machine is using. I'd also recommend updating the libraries (especially transformers, accelerate, and bitsandbytes), it might solve a compatibility issue.
i tried running the same notebook on gcp A100 machine, and it failed on :
`File ~/.local/lib/python3.9/site-packages/transformers/utils/bitsandbytes.py:109, in set_module_quantized_tensor_to_device(module, tensor_name, device, value, fp16_statistics)
107 new_value = old_value.to(device)
108 elif isinstance(value, torch.Tensor):
--> 109 new_value = value.to(device)
110 else:
111 new_value = torch.tensor(value, device=device)
RuntimeError: CUDA error: out of memory
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
Compile with
TORCH_USE_CUDA_DSA
to enable device-side assertions.`on colab it work perfectly.
any idea ?
The text was updated successfully, but these errors were encountered: