Haruka1307

Results 9 comments of Haruka1307

RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:6 and cuda:0!

> > Could you try `export CUDA_VISIBLE_DEVICES=6` instead? > > It works for 80G momory devices but not for less memory devices. Could you please share any advice for other...

SFT Mistral；

I also met this err... would you provide more details about your environment ?

[Bug] DeepSeek R1 32B 模型 测评 AIME2024 数据集 得分低

> configs/datasets/aime2024/，你用哪个版本的数据集就改哪个版本的代码，改max_out_len 想问一下这是要改site-packages里的代码吗

ValueError: Trying to set a tensor of shape torch.Size([32769536]) in "weight" (which has shape torch.Size([32001, 4096])), this looks incorrect.

I met the same situation.

ValueError: Trying to set a tensor of shape torch.Size([32769536]) in "weight" (which has shape torch.Size([32001, 4096])), this looks incorrect.

> I met the same situation. how about downgrade Accelerate to 0.30.0?

ValueError: Trying to set a tensor of shape torch.Size([32769536]) in "weight" (which has shape torch.Size([32001, 4096])), this looks incorrect.

I finally sovle this err by upgrading python from 3.9->3.10,without downgrading accelerate.

MATH dataset unavailable

https://www.modelscope.cn/datasets/opencompass/competition_math/summary

TRAK on cuda version 11.8 have numerical issue

I encounter the same error

TRAK on cuda version 11.8 have numerical issue

> I encounter the same error Finally I use Basicprojector, it almost has the same speed