openai

/

gpt-oss-120b

like 3.5k

Follow

OpenAI 17.9k

Hello,

thanks for your great work!

When I try to reproduce the results on hle, I get a much lower acc as 8%, which is 14.9% from your blog.

1. My server is lanuched with vllm

vllm serve gpt-oss-120b

2. I use the code from

https://github.com/centerforaisafety/hle

in which the temperature and top_p are not set. May this be the key problem?

And are there any other points I need to pay attention to?

Best regards!