Hacker News Reader

Comment by rdli | Hacker News Reader

I'm working on something like this! It's simple in concept, but there are lots of fiddly bits. A big one is performance (at least, without spending $$$$$ on GPUs.) I haven't found that much in terms of how to tune/deploy LLMs on commodity cloud hardware, which is what I'm trying this out on.

View on HN

You can use ONXX versions of embedding models. Those run faster on CPU.

Also, don’t discount plain old BM25 and fastText. For many queries, keyword or bag-of-words based search works just as well as fancy 1536 dim vectors.

You can also do things like tokenize your text using the tokenizer that GPT-4 uses (via tiktoken for instance) and then index those tokens instead of words in BM25.

Thanks! I should have been clearer -- embeddings are pretty fast (relatively) -- it's inference that's slow (I'm at 5 tokens/second on AKS).

Could you sidestep inference altogether? Just return the top N results by cosine similarity (or full text search) and let the user find what they need?

https://ollama.com models also works really well on most modern hardware

I'm running ollama, but it's still slow (it's actually quite fast on my M2). My working theory is that with standard cloud VMs, memory <-> CPU bandwidth is an issue. I'm looking into vLLM.

And as to sidestepping inference, I can totally do that. But I think it's so much better to be able to ask the LLM a question, run a vector similarity search to pull relevant content, and then have the LLM summarize this all in a way that answers my question.

Oh yeah! What I meant is having Ollama run on the user's machine. Might not work for the use case you're trying to build for though :)

This style of embeddings could be quite lightweight/cheap/efficient https://github.com/cohere-ai/BinaryVectorDB

Embedding models are generally lightweight enough to run on CPU, can be done in the background while the user isn't using their device.

Reply on news.ycombinator.com