LocalLLaMA

2739 readers

52 users here now

Welcome to LocalLLama! This is a community to discuss local large language models such as LLama, Deepseek, Mistral, and Qwen.

Get support from the community! Ask questions, share prompts, discuss benchmarks, get hyped at the latest and greatest model releases! Enjoy talking about our awesome hobby.

As ambassadors of the self-hosting machine learning community, we strive to support eachother and share our enthusiasm in a positive constructive way.

founded 2 years ago

MODERATORS

SkySyrup@sh.itjust.works

pax@sh.itjust.works

noneabove1182@sh.itjust.works

Smokeydope@lemmy.world

MonsterBug@sh.itjust.works

Recommendations for a lightweight Python LLM framework for a webapp? (palaver.p3x.de)

submitted 5 days ago* (last edited 5 days ago) by hendrik@palaver.p3x.de to c/localllama@sh.itjust.works

5 comments fedilink hide all child comments

I'm developing a small Python webapp as some sort of finger exercise. Mostly a chatbot. I'm using the Quart framework, which is pretty much alike Flask, just async. Now I want to connect that to a LLM inference endpoint. And while I could do the HTTP requests myself, I'd prefer something that does that for me. It should support the usual OpenAI style API, in the end I'd like it to connect to things like Ollama and KoboldCPP. No harm if it supports image generation, agents, tools, vector databases, but that's optional.

I've tried Langchain, but I don't think I like it very much. Are there other Python frameworks out there? What do you like? I'd prefer something relatively lightweigt that gets out of the way. Ideally provider agnostic, but I'm mainly looking for local solutions like the ones I mentioned.

Edit: Maybe something that also connects to a Runpod endpoint, to do inference on demand (later on)? Or at least something which I can adapt to that?

top 5 comments

sorted by: hot top controversial new old

[–] Oskar@piefed.social 3 points 4 days ago

Ollama's Python API works well and there's a lot of examples. However, I've not gotten the Ollama REST API to work, the response doesn't unpack into json for me.

Just design the rest of your system so that it doesn't have to know anything about the implementation (only prompts and responses) and you should be able to easily replace the LLM part later.

[–] ryedaft@sh.itjust.works 2 points 5 days ago (1 children)

Try Simon Willison's LLM library. Python and command line connection to LLMs, local or remote. Lots of blog posts about how to do various things. Also image examples.

https://github.com/simonw/llm

[–] hendrik@palaver.p3x.de 2 points 5 days ago

Thanks to all who sent suggestions, I think I'll try this one first. Looks quite neat and ticks all the boxes. If the async plugin does it's job, I can use that.

[–] gregs_gumption@lemm.ee 2 points 5 days ago

Llamafile would probably meet your needs.

[–] bjoern_tantau@swg-empire.de 2 points 5 days ago

Ollama has their own Python library linked on their website.