This update is pretty exciting, I’m gonna try running a large model (65B) with a 3090. I have ran a ton of local LLM but the hardest part is finding out the prompt structure. I wish there is some sort of centralized data base that explains it.
I tried llama-65b on a system with RTX 4090 + 64Gb of DDR5 system RAM. I can push up to 45 layers (out of 80) to the GPU, and the overall performance is ~800ms / token, which is "good enough" for real-time chat.
Supposedly the new update with GPU offloading will bring that up to 10 tokens per second! 1 token per second is painfully slow, that’s about 30s for a sentence.
