Overriding torch_dtype=None with `torch_dtype=torch.float16` due to requirements of `bitsandbytes` to enable model loading in mixed int8. Either pass torch_dtype=torch.float16 or don't pass this argument at all to remove this warning. - Githubissues

ssbuild / chatglm_finetuning

chatglm 6b finetuning and alpaca finetuning

1.53k stars 175 forks source link

Overriding torch_dtype=None with `torch_dtype=torch.float16` due to requirements of `bitsandbytes` to enable model loading in mixed int8. Either pass torch_dtype=torch.float16 or don't pass this argument at all to remove this warning. #235

Open sanwei111 opened 1 year ago

sanwei111 commented 1 year ago

我用in8，需要改torch_dtype参数吗