Re:Code Llama, an AI tool for coding
by int_19h on Thursday August 24, 2023 @03:12PM
(#37253275)
12Gb of VRAM lets you run 13B models (4-bit quantized) with reasonable speed, and can be had for under $300 if you go for previous-generation NVidia hardware. Plenty of developers around with M1 and M2 Macs, as well.
[ Parent
| on HN ]
There are no comments on this comment yet.