在使用subjective数据集时报错
#1459
Replies: 1 comment 1 reply
-
|
客观和主观的评测方式不一样，不能放在一起跑的，主观推荐用config的方式跑 |
Beta Was this translation helpful? Give feedback.
1 reply
-
好的，明白了，多谢 |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
python run.py
--models hf_qwen2_1_5b_instruct
-datasets demo_gsm8k_chat_gen demo_math_chat_gen alignbench_judgeby_autoj
--debug
-a vllm
我在跑数据集alignbench_judgeby_autoj或者mtbench时，在eval阶段报错TypeError: LMEvaluator.init() missing 2 required positional arguments: 'judge_cfg' and 'output_path'。请问这个是什么原因，该怎么修改？
Beta Was this translation helpful? Give feedback.
All reactions