Stop sequences parameter in gpt.py should be None?

[zrquan]

When I use the OpenAI Chat Completions API provided by Ollama, the content returned by LLM is empty. And when the stop parameter is set to None, LLM can respond normally.

[clairew]

Hey, appreciate the feedback. We didn't test the framework with Ollama. Infrastructure for using models is ultimately up to the user - we simply provided the model wrappers we used in our experiments as an example.

If you'd like - you can make a pull request to include generalized Ollama support! Thanks.