search

anarchy-ai / LLM-VM

irresponsible innovation. Try now at https://chat.dev/
https://anarchy.ai/
MIT License
477 stars 149 forks source link

Support fine-tuning for chatgpt 3.5 #132

Open INF800 opened 1 year ago

INF800 commented 1 year ago

We can fine-tune gpt 3.5 i.e chat_gpt according to official blog and docs

So, isn't it a good idea to remove the below line and make sure fine-tuning happens? https://github.com/anarchy-ai/LLM-VM/blob/3b4ade5a40c5160b71642373ae19ba8391066f84/src/llm_vm/onsite_llm.py#L444

Note: gpt-3.5-turbo-0301 is used in generate method but gpt-3.5-turbo-0613 supports fine-tuning. Suffix is a different number.

INF800 commented 1 year ago

Can send a PR for this if required. Please let me know.

mmirman commented 1 year ago

It's probably quite a bit more than just removing that line, but this is a good first issue! I think 3.5 has different formats than 3 so the PR should have tests in it :-) But thanks for submitting the issue!

INF800 commented 1 year ago

Makes sense. Will make required changes and send PR with tests. I have a small doubt - do you use pytest for writing tests or some other testing tool? Asking because I don't see pytest being used in tests/ directory.

mmirman commented 1 year ago

No pytest yet - best to make a ticket for it!

INF800 commented 1 year ago

No pytest yet - best to make a ticket for it!

Opened #133

bilal-aamer commented 9 months ago

Will attempt to close this