Comment by ac130kz

Comment by ac130kz 3 days 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 a day 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 3 days 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