2 ms·
I'm happy that someone implemented a CPU-only version of Llama. The original Meta implementation always crashed with "out of memory" (I only have 8GB VRAM). Too
by KETpXDDzR 4y ago
I'm happy that someone implemented a CPU-only version of Llama. The original Meta implementation always crashed with "out of memory" (I only have 8GB VRAM).
Too bad Llama doesn't work as well as (Chat-)GPT for instructional prompts.