Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why we no longer use LangChain for building our AI agents

octomind.dev

Post reply on HN
Print view View original

151–160 of 307 posts

Re: Why we no longer use LangChain for building our AI agents

#151

Post by wongarsu »

moffkalast wrote: post #134
netdevnet wrote: post #124

Earlier quoted context omitted.

Chat GPT is just GPT version 3.5. OpenAI released many other versions of GPT before that. In fact, Open AI became really popular around the time of the GPT 2 which was a fairly good chat model. Also, the Transformer architecture was not created by OpenAI so LLMs were a thing way before OpenAI existed :)

GPT-2 was not a fairly good chat model, it was a completely incoherent completion model. GPT-3 was not much better overall (take any entry level 1B sized model you can find today and it'll steamroll it in every way, hell probably even smaller ones), and the public at large never really had any access to it, I vaguely recall GPT 3 being locked behind an approval only paid API or something unfeasible like that. Nobody…

OpenAI had a real issue with making (for their time) great models but streching their rollout over months. They gave access to press and some twitter users, everyone else had to apply for their use case only to be put on the waitlist. That completely killed any momentum.

The first version of ChatGPT wasn't a huge leap from simulating chat with instruction-tuned GPT 3.5, the real innovation was scaling it to the point where they could give the world immediate and free access. That built the hype, and that success allowed them to make future ChatGPT versions a lot better than the instruction-tuned models ever were.

Re: Why we no longer use LangChain for building our AI agents

#152

Post by netdevnet »

moffkalast wrote: post #134
netdevnet wrote: post #124

Earlier quoted context omitted.

Chat GPT is just GPT version 3.5. OpenAI released many other versions of GPT before that. In fact, Open AI became really popular around the time of the GPT 2 which was a fairly good chat model. Also, the Transformer architecture was not created by OpenAI so LLMs were a thing way before OpenAI existed :)

GPT-2 was not a fairly good chat model, it was a completely incoherent completion model. GPT-3 was not much better overall (take any entry level 1B sized model you can find today and it'll steamroll it in every way, hell probably even smaller ones), and the public at large never really had any access to it, I vaguely recall GPT 3 being locked behind an approval only paid API or something unfeasible like that. Nobody…

You are saying that after having experienced all the subsequent versions. GPT-2 was fairly good, not impressive but fairly good. People were using for all sorts of stuff for the fun of it. The GPT 3 versions were really impressive and had everyone here super excited

Re: Why we no longer use LangChain for building our AI agents

#154

Post by 9dev »

dartos wrote: post #138
9dev wrote: post #137

Earlier quoted context omitted.

Well. I'm working on a product that relies on both AI assistants in the user-facing parts, as well as LLM inference in the data processing pipeline. If we let our LLM guy run free, he would create an inscrutable tangled mess of Python code, notebooks, Celery tasks, and expensive VMs in the cloud. I know Pythonista's regard themselves more as artists than engineers, but the rest of us needs reliable and deterministica…

Sounds like your LLM guy just isn’t very good. Langchain is, when you boil it down, an abstraction over text concatenation, staged calls to open ai, and calls to vector search libraries. Even without standard tooling, an experienced programmer should be able to write an understandable system that does those things.

Fair point. The overlap of machine learning savvy, experienced engineer, and ready to work for a startup's salary in Germany just isn't too big.

Re: Why we no longer use LangChain for building our AI agents

#155

Post by blackkettle »

fforflo wrote: post #133

LLM frameworks like LangChain are causing a java-fication or Python . Do you want a banana? You should first create the universe and the jungle and use dependency injection to provide every tree one at a time, then create the monkey that will grab and eat the banana.

Holy moly this was _exactly_ my impression. It seems to really be proliferating and it drives me nuts. It makes it almost impossible to useful things, which never used to be a problem with Python - even in the case of complex projects.

Figuring out how to customize something in a project like LangChain is positively Byzantine.

Re: Why we no longer use LangChain for building our AI agents

#156

Post by sabbaticaldev »

fforflo wrote: post #133

LLM frameworks like LangChain are causing a java-fication or Python . Do you want a banana? You should first create the universe and the jungle and use dependency injection to provide every tree one at a time, then create the monkey that will grab and eat the banana.

I’ll use this to explain why typescript is bad

Re: Why we no longer use LangChain for building our AI agents

#157

Post by moffkalast »

netdevnet wrote: post #152
moffkalast wrote: post #134

Earlier quoted context omitted.

GPT-2 was not a fairly good chat model, it was a completely incoherent completion model. GPT-3 was not much better overall (take any entry level 1B sized model you can find today and it'll steamroll it in every way, hell probably even smaller ones), and the public at large never really had any access to it, I vaguely recall GPT 3 being locked behind an approval only paid API or something unfeasible like that. Nobody…

You are saying that after having experienced all the subsequent versions. GPT-2 was fairly good, not impressive but fairly good. People were using for all sorts of stuff for the fun of it. The GPT 3 versions were really impressive and had everyone here super excited

I'd argue the GPT-3 results were really cherry picked by the few people who had access, at least if the old versions of 3.5 and turbo are anything to go by. The hype would've died instantly if anyone had actually tried them themselves and realized that there's no consistency.

If you want to try out GPT-2 to refresh your memory, here [0] is an online demo. It's bad, I'd say worse than classical graph/tree based autocomplete. I'm fairly sure Swiftkey makes more coherent sentences.

[0] https://transformer.huggingface.co/doc/gpt2-large

Re: Why we no longer use LangChain for building our AI agents

#158

Post by captaincaveman »

I think LangChain basically tried to do a land grab, insert itself between developers and LLM's. But it didn't add significant value and seemed to dress it up by adding abstractions that didn't really make sense. It was that abstraction gobbledygook smell that made me cautious.

Re: Why we no longer use LangChain for building our AI agents

#159

Post by d4rkp4ttern »

Frustration with LangChain is what led us (ex-CMU/UW-Madison researchers) to start building Langroid[1], a multi-agent LLM framework. We have been thoughtful about designing the right primitives and abstractions to enable a simple developer experience while supporting sophisticated workflows using single or multiple agents. There is an underlying loop-based orchestration mechanism that handles user interaction, tool handling and inter-agent handoff/communication.

We have companies using Langroid in production.

[1] Langroid: https://github.com/langroid/langroid

Re: Why we no longer use LangChain for building our AI agents

#160

Post by codelion »

Many such cases. It is very hard to balance composition and abstraction in such frameworks and libraries. And LLMs being so new it has taken several iterations to get the right patterns and architecture while building LLM based apps. With patchwork (https://github.com/patched-codes/patchwork) an open-source framework for automating development workflows we try hard to avoid it by not abstracting unless we see some client usage. As a result you do see some workflows appear longer with many steps but it makes it easier to compose them.
Post reply on HN