3 ms·
With 32 GB RAM you can do inference with quantized 34b models. I wouldn’t call that useless? You don’t need a GPU for llm inference. Might not be as fast as it
by tosh 3y ago
With 32 GB RAM you can do inference with quantized 34b models. I wouldn’t call that useless?
You don’t need a GPU for llm inference. Might not be as fast as it could be but usable.