how to use a local LLM to evaluate prediction quality? For example, Llama-3-70B-Instruct?

[txchen-USTC]

Feature request / 功能建议

how to use a local LLM to evaluate prediction quality? For example, Llama-3-70B-Instruct?

Motivation / 动机

how to use a local LLM to evaluate prediction quality? For example, Llama-3-70B-Instruct?

Your contribution / 您的贡献

how to use a local LLM to evaluate prediction quality? For example, Llama-3-70B-Instruct?

[bys0318]

Hi, you can use the evaluation code in eval_quality.py to evaluate the generation quality. Substitute the API reference call get_response_gpt4 with your local LLM model call.