When PR #92 gets merged it will be possible to use Ollama to host local LLM's that will be used as a backend for yai. Smaller models that require less computing power are sometimes confused by built-in prompts meant for gpt-3.5-turbo. Allowing to set custom prompts in the config file means people will be able to use prompts that are best for their use case.
When PR #92 gets merged it will be possible to use Ollama to host local LLM's that will be used as a backend for yai. Smaller models that require less computing power are sometimes confused by built-in prompts meant for gpt-3.5-turbo. Allowing to set custom prompts in the config file means people will be able to use prompts that are best for their use case.