3 ms·
Yup. The problem with frameworks is they assume (historically mostly but not always correctly) that layers of abstraction mean one can forget about the layers b
by danielmarkbruce 2y ago
Yup. The problem with frameworks is they assume (historically mostly but not always correctly) that layers of abstraction mean one can forget about the layers below. This just doesn't work with LLMs. The systems are closer to biology or something.
- nosefurhairdo 2y agoVery much depends on the framework. I'm currently building a GitHub App with the Probot framework, which mostly just handles authentication boilerplate and some testing niceties, then just gives you an authenticated GitHub API client (no facade/abstraction). Then of course there's the many web application frameworks, because nobody in their right mind would want to implement http request parsing themselves (outside of academic exercises). In fact, I would argue that most popular frameworks exist precisely because it's often more time efficient to forget about underlying details. All computer software is built on abstraction. The key is picking the right level of abstraction for your use case.
- danielmarkbruce 2y agoReread the thread and the comment. It's about the LLM frameworks and acknowledges that most non LLM frameworks historically are helpful and correct in abstracting away details.
- nosefurhairdo 2y agoAh the bit in parentheses was worded such that I misunderstood your point.
- danielmarkbruce 2y agoThat bit is poorly worded. I should have had a comma before the last word. My bad.
- randomdata 2y agoIt often took quite a long time for those historic frameworks to get the abstraction right. Survivorship bias sees us forget all the failed attempts. I'm unconvinced there is no room for a framework here because LLMs are somehow special. LangChain just missed the mark. Unsurprisingly so, it being an early attempt, not to mention predating general availability of the LLM chatbots that have come to define the landscape.