CoolFace
Modelpublic

minjaechoi/nemotron3-nano-30b-a3b-2p25bit-r17

sourceHugging Faceupdated 8d agoView on Hugging Face
0likes268downloads
Model Card

NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 — 2.254-bit routed experts (r17)

Internal research checkpoint. Routed experts average 2.254 bits; every other weight is BF16. Weights are stored dequantized in BF16 tensors and load with stock transformers / vLLM.

Base modelnvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
Routed-expert average2.254 bits
Internal IDr17

License follows the base model.