Kind request for updating MT-Bench leaderboards with Qwen1.5-Chat series

[Lukeming-tsinghua]

Hi LM-Sys team, we would like to present the generation results and self-report scores of Qwen1.5-7B-Chat, Qwen1.5-14B-Chat, and Qwen1.5-72B-Chat on MT-Bench. Could you kindly help us verify them and update the leaderboard to add the Qwen2-beta-Chat series? The generation results are attached in the zip file.

qwen2-beta-mtbench-results.zip

Model	Self-report Score
Qwen1.5-72B-Chat	8.67
Qwen1.5-14B-Chat	7.99
Qwen1.5-7B-Chat	7.58

[infwinston]

thanks! is qwen2-beta publicly available now?

[Lukeming-tsinghua]

Thanks for your help!

We plan to make the HF repos public and begin to provide APIs today. By the way, we use the official name Qwen1.5-Chat instead of the previous Qwen2-beta in the zip file. Sorry for the confusion we made.

[OpenJarvisAI]

@Lukeming-tsinghua Does there only same 7b 14b 70b in Qwen1.5?

[gblazex]

great work guys!
will this be up on Human Arena as well? 🤞👀

[infwinston]

yes we're testing it out. will soon launch.

[gblazex]

this can be closed i think