RedHatAI/diffusiongemma-26B-A4B-it-FP8-dynamic

Quantization vs google/diffusiongemma-26B-A4B-it. RedHatAI/diffusiongemma-26B-A4B-it-FP8-dynamic is a quantized version of the other (F8_E4M3/BF16).

On the Hub · Downloads on Model Pulse