feat: use google genai

fix: hardcode the model max length for vllm

fix: remove vllm max length

remove check_gt_only flag

fix: optional bool args

fix passk

Merge pull request #77 from shwinshaker/add-unique-cache

Specify a unique cache directory before each code execution

Merge pull request #75 from zhangchen-xu/main

fix make_raw_chat_prompt when prefill is disabled

fix prefill arg

update doc