nvidia/diffusiongemma-26B-A4B-it-NVFP4

Quantization vs google/diffusiongemma-26B-A4B-it. nvidia/diffusiongemma-26B-A4B-it-NVFP4 is a quantized version of the other (U8/BF16).

3 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse