NVIDIA / NVIDIA/Model-Optimizer
MiniMax M3 nvfp4 ckpt
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 3.8k
- Forks
- 604
- Avg merge
- 2d 8h
- Merged PRs (30d)
- 142
Description
One of the main features of NVIDIA latest flagship Blackwell GPUs is NVFP4, the ability to have better perf at similar accuracy to BF16 due to having block size 16 and using E4M3 as the block scale instead of E8M0 unlike MXFP4.
what is the timelines for nvfp4 ckpts for MiniMax M3 will be out? https://huggingface.co/nvidia/models?search=nvfp4
OpenRouter (which hobbysts and small enterepises) use consistently show that MiniMax M3 for the past weeks consistently ranks high in popularity
+viz @xinli-sw @ncomly-nvidia @kedarpotdar-nv @mxinO @Edwardf0t1
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No source file, test, or implementation entry point is identified in the issue. Start by reviewing the Model-Optimizer repository and the linked NVIDIA Hugging Face model search; the request is complete when NVFP4 MiniMax M3 checkpoint availability and its timeline are established.
Written by the indexing model from the issue text.
Assessment
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100