I also get similar result. performace slows when mtp is enabled for qwen3.5-122b-a10b-nvfp4
Any instructions for NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4?
Or Australian…it’s been in common use here for at least 80 years.
I also get similar result. performace slows when mtp is enabled for qwen3.5-122b-a10b-nvfp4
Any instructions for NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4?
Or Australian…it’s been in common use here for at least 80 years.