nintwentydo/pixtral-12b-FP8-dynamic-FP8-KV-cache

Quantization vs mistral-experimental/pixtral-12b. nintwentydo/pixtral-12b-FP8-dynamic-FP8-KV-cache is a quantized version of the other (F8_E4M3/BF16).

On the Hub · Downloads on Model Pulse