Ugh, I feel exactly the same way. In fact, I've spent the last 2 weeks refactoring our entire LLM pipeline to completely root out LangChain. I'm just implementing basic RAG from scratch, which 1) has me learning a lot more quickly and 2) feeling much more in control of the entire pipeline.
LangChain Is a Black Box
21–30 of 31 posts
Re: LangChain Is a Black Box
#22I don't know if this will help you, but have you tried langroid?
Even if you don't want to use agents, it is still useful as a convenient library for calling an Open AI compatible endpoint.
https://langroid.github.io/langroid/quick-start/llm-interact...
Re: LangChain Is a Black Box
#23I don't know if this will help you, but have you tried langroid? Even if you don't want to use agents, it is still useful as a convenient library for calling an Open AI compatible endpoint. https://langroid.github.io/langroid/quick-start/llm-interact...
Thanks for mentioning Langroid[1], I am the lead architect/dev. We are starting to see companies use it in production after evaluating alternative frameworks, whether agent-oriented or not.
We started building Langroid in early 2023 after finding existing frameworks lacking in terms of good dev-ex, extensibility and clarity of code. We prioritize code transparency, flexibility, stability, good test coverage. We designed it to be agent-oriented from the start, with an elegant agent orchestration mechanism loosely inspired by various paradigms such as Blackboard Architectures, Actor Model, and Production systems, and process calculi. But as you said, Langroid is useful even if you want to just use a single LLM conversation-state.
Re: LangChain Is a Black Box
#24[flagged]
Re: LangChain Is a Black Box
#25[flagged]
Is this written by a human?
Re: LangChain Is a Black Box
#26Ugh, I feel exactly the same way. In fact, I've spent the last 2 weeks refactoring our entire LLM pipeline to completely root out LangChain. I'm just implementing basic RAG from scratch, which 1) has me learning a lot more quickly and 2) feeling much more in control of the entire pipeline.
How are you running the pipeline exactly? I'm new to AI dev and when I hear pipeline I think either CI/CD or something like Apache Airflow or Kestra.
Re: LangChain Is a Black Box
#27Howdy! Erick from LangChain here. Would love to learn a bit more about your bug hunting process and see how we can improve the debugging process for everyone else! Could you email me at erick@langchain.dev?
Erick, please just read this entire thread.
Re: LangChain Is a Black Box
#28Oh my god, I feel exactly the same way. I am not a programmer, but I'm excited by the potential of LLMs in my work. So, I've dusted off my python skills that I haven't used since I was in an fMRI lab, and experimenting to see what works. First of all, every library/framework I've found is moving so fast that all the tutorials and printed material (O'Reilly books etc) are already out of date. Many of the changes are o…
Just in case anyone comes back to this, I found the following resource helpful:
Brandon Hancock’s (3 hour!!) YouTube video and accompanying GitHub repo: https://github.com/bhancockio/langchain-crash-course https://youtu.be/yF9kGESAi3M
I’m not in a position to say, but perhaps many people will ultimately go with directly using the python libraries and APIs of their components. But if you are messing around with different components until you settle on something, it’s useful.
Re: LangChain Is a Black Box
#29We're moving our entire codebase away from Langchain and related products (langgraph, langsmith). I originally thought the library had potential and just needed some polishing which would come with time, but it's just been getting worse and worse. LCEL is the weirdest half-baked idea; it's a nice gimmick that makes your code impossible to debug in a step through debugger. Random stuff breaks with every other release.…
Are you replacing LangGraph with something else? I've luckily avoided LangChain exactly because it's a black box, but been on the lookout for some kind of DAG visualization of LLM flows. LangGraph pops up first on any related searches.
We're not sure what to do with langgraph yet, interested if anyone here has tried viable alternatives. We could implement a simplified execution graph in-house, but it feels like reinventing the wheel. Also thinking to keep the "graph" part of langgraph but replace all everything else (i.e. tool calls, retrievers, etc.) by non-langchain alternatives.
Re: LangChain Is a Black Box
#30Just in case anyone comes back to see this: I found the following (3 hour!!) YouTube video to be really helpful. It starts slow, but skip ahead and watch in 2x until you get to what you need.