Comment by ac130kz

Comment by ac130kz 10 months ago

2 replies

I've tried running 3B LLAMA on my old machine and it was sluggish and kinda bad, but Claude 3.5 Sonnet works great. I will try the 70B model, should be good enough.

replete 10 months ago

I didn't have much success with ollama either. You really need a great accelerator for it to be useful enough. Codestral however is working out great so thanks for the tip.

dartos 10 months ago

Yeah, I’ve found local models are completely useless for coding tasks whereas openai and Claude can at least do some refactors with guidance