New ask Hacker News story: ChatGPT is 1yo now. Which LLM framework(s) do you use in production and why?
ChatGPT is 1yo now. Which LLM framework(s) do you use in production and why?
2 by behnamoh | 0 comments on Hacker News.
I've come across many LLM frameworks: Langchain, LlamaIndex, LMQL, guidance, Marvin, Instructor, etc. There's a lot of overlap between them and I don't know if any of them actually adds a value to LLM workflows in a way that's maintainable and robust. So far, I've been able to just build my own little libraries to use in some LLM applications (no RAG), but as I consider the more recent advancements in the field (guaranteed function calling, better RAG, agents and tool use, etc.), I wonder if using one of these frameworks would be a better approach compared to building everything on my own. I appreciate your thoughts and comments on this!
2 by behnamoh | 0 comments on Hacker News.
I've come across many LLM frameworks: Langchain, LlamaIndex, LMQL, guidance, Marvin, Instructor, etc. There's a lot of overlap between them and I don't know if any of them actually adds a value to LLM workflows in a way that's maintainable and robust. So far, I've been able to just build my own little libraries to use in some LLM applications (no RAG), but as I consider the more recent advancements in the field (guaranteed function calling, better RAG, agents and tool use, etc.), I wonder if using one of these frameworks would be a better approach compared to building everything on my own. I appreciate your thoughts and comments on this!
Comments
Post a Comment