41–47 of 47 posts
Earlier quoted context omitted.
If their native language isn’t English then I’ll come off my pedestal
Still, there is a big difference between "Translate my answer into English: " and "Answer this question: ".
Earlier quoted context omitted.
Do not buy a 4 GB card for this. Mine is an RTX 3050 Laptop, I picked it because it is boring hardware that many people already have. If you are buying, buy VRAM. At 0.5B where I could measure both, resident training was 1.43x faster than streaming. And if you do stream, system RAM matters more, the base sits there and has to page-lock. 16 GB is about the floor for 8B.
what price is the cheapest
Earlier quoted context omitted.
Do not buy a 4 GB card for this. Mine is an RTX 3050 Laptop, I picked it because it is boring hardware that many people already have. If you are buying, buy VRAM. At 0.5B where I could measure both, resident training was 1.43x faster than streaming. And if you do stream, system RAM matters more, the base sits there and has to page-lock. 16 GB is about the floor for 8B.
what price is the cheapest
the cheapest one is the one you already have.
more vram for the money is what i'd look for. usually used :)
Author here. The constraint everyone works around is that the frozen base has to fit in VRAM. But during LoRA the base is frozen — read, never written. It doesn't need to live in VRAM, it needs to arrive before the matmul that uses it. So it sits in host RAM and streams into a small pool of pre-allocated VRAM buffers, one decoder layer at a time, prefetched one ahead on a dedicated CUDA stream. Peak VRAM becomes one…