You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is fine right now because the only functionality really being used is chat completions and embedding generation. However, if we add multi-modal (image) support, tool support, etc, each provider could have its own weird differences that turn into a pain to maintain.
I'm hoping to find a good javascript/typescript library to abstract most of that complexity out. These are some of the options:
Other than using a library, another option talked about before is to only support openai/ollama and then rely on an external proxy/router like litellm that does the translation to other llm provider apis.
The text was updated successfully, but these errors were encountered:
This issue is mostly so I don't forget about this and can write my findings somewhere.
Currently, we have interfaces for different LLMs from scratch in https://github.com/justyns/silverbullet-ai/tree/main/src/providers
This is fine right now because the only functionality really being used is chat completions and embedding generation. However, if we add multi-modal (image) support, tool support, etc, each provider could have its own weird differences that turn into a pain to maintain.
I'm hoping to find a good javascript/typescript library to abstract most of that complexity out. These are some of the options:
langchain.jshopfieldhuggingface.jsOther than using a library, another option talked about before is to only support openai/ollama and then rely on an external proxy/router like litellm that does the translation to other llm provider apis.
The text was updated successfully, but these errors were encountered: