Humaneval, use base model or instruct finetuned model?

deepseek-ai / DeepSeek-LLM

DeepSeek LLM: Let there be answers

https://chat.deepseek.com/

MIT License

1.47k stars 94 forks source link

Open jasonzliang opened 8 months ago

jasonzliang commented 8 months ago

Which one would you recommend to use to get the best possible performance?

DeepSeekPH commented 7 months ago

The finetuned model yields better Humaneval results.