Alibaba-NLP

/

gme-Qwen2-VL-2B-Instruct

like 135

Follow

Alibaba-NLP 1.65k

Hi, could you please provide the version of your python environment?

I used the following versions:

• python 3.8 / 3.11
• torch 2.5.1+cu118
• transformers 4.45.0.dev0 / transformers 4.46.3

I noticed that my inference results differ from the ones mentioned in your comment:

• My results: tensor([0.2747, 0.5591], dtype=torch.float16) vs. Comment: tensor([0.2281, 0.6001], dtype=torch.float16)
• My results: tensor([0.3040, 0.6646], dtype=torch.float16) vs. Comment: tensor([0.2433, 0.7051], dtype=torch.float16)
• My results: tensor(0.6606, dtype=torch.float16) vs. Comment: tensor(0.6108, dtype=torch.float16)

I’m concerned that there may be an error due to the discrepancies between my results and the ones you provided. Do you have any suggestions?