Quantization publisher
nvidia
nvidia publishes 1 quantizations across 1 models in our index, averaging 5.712 effective bits per weight.
From the file· summed file bytes
Repositories
1
Quantizations
1
Models covered
1
Avg effective bpw
5.712
across their files
Models they publish
| Model | Quantizations | Smallest |
|---|---|---|
| NVIDIA-Nemotron-3-Nano-4B-FP8 | 1 | 2.64 GiB |