search

QwenLM / Qwen2

Qwen2 is the large language model series developed by Qwen team, Alibaba Cloud.
7.04k stars 416 forks source link

how to reproduce QWEN1.5-7B-CHAT results #382

Closed chunniunai220ml closed 1 month ago

chunniunai220ml commented 3 months ago

how to reproduce QWEN1.5-7B-CHAT results as you report:

image

i got to TOEFL=30.198 by https://github.com/OpenLMLab/LEval as u mentioned

hzhwcmhf commented 3 months ago

tpo.pred.json

Here is the output. We use greedy decoding, and the score may vary a little due to the precision problem. @chunniunai220ml

chunniunai220ml commented 3 months ago

tpo.pred.json

Here is the output. We use greedy decoding, and the score may vary a little due to the precision problem. @chunniunai220ml

can u provide example code with LEVAL? like https://github.com/OpenLMLab/LEval/blob/main/Baselines/llama2-chat-test.py, i modified the file to adapt QWEN1.5-CHAT-7B, but failed to reproduce your results

chunniunai220ml commented 2 months ago

@hzhwcmhf hi, any further infomation for example code with LEVAL

github-actions[bot] commented 1 month ago

This issue has been automatically marked as inactive due to lack of recent activity. Should you believe it remains unresolved and warrants attention, kindly leave a comment on this thread.