TypeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

[ArtificialZeng]

Traceback (most recent call last): File "/home/zengzijian/ai_code/fastllm/build/tools/baichuan2flm.py", line 18, in torch2flm.tofile(exportPath, model, tokenizer, dtype = dtype) File "/home/zengzijian/ai_code/fastllm/build/tools/fastllm_pytools/torch2flm.py", line 156, in tofile cur = dict[key].numpy().astype(ori_np_data_type) TypeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.

[ztxz16]

这个有点奇怪，有改过脚本吗？baichuan的脚本里11行已经model.to("cpu")，为什么这里还是在cuda:0上

[TylunasLi]

正常的baichuan转换如楼上所说是做过model.to("cpu")的，其他模型，如果model.to("cpu")不成功，可以在这里加一下，改为：

dict[key].cpu().numpy().astype(ori_np_data_type)