Quantization vs google/gemma-4-12B-it. AxionML/Gemma-4-12B-NVFP4 is a quantized version of the other (BF16/U8).
On the Hub · Downloads on Model Pulse