3 ms·
When I was creating this tool, I made sure to abstract out the reliance on just one LLM or vector db. Instead, I focused on using langchain/huggingface for toke
by raghavankl 3y ago
When I was creating this tool, I made sure to abstract out the reliance on just one LLM or vector db. Instead, I focused on using langchain/huggingface for tokenization, embedding, and conversational modeling. This was done purposely so that it would be simple to replace the OpenAI dependencies with any other LLMs if needed.