nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16

Update README.md

Switch vision encoder reference from C-RADIOv2-H to C-RADIOv4-H; remove now-redundant _patch_cpe_align_corners (#10)

Remove vLLM NVFP4 tp>1 caveat (#11)

Fix rendering

Remove name from model card

Update model card gpu requirements

Fix pipeline tag

Fix pipeline tag

Add citation to README (#6)

Add dataset tags

Update README.md

Add hyperlinks

Fix rendering problems

Fix rendering problem

Add YAML frontmatter to model card

Fix "download the weights" rendering (#4)

Fix an inconsistency

Rename overview.md to README.md so HF renders the model card

Add model card

Remove quick tests

Add media dir

Enable batched audio inference (#3)

Support Transformers 5.6, dynamic-resolution images, temporal video embedder, vLLM parity (#2)

Update chat template

Revert modeling_nemotron_h.py

Update generation_config with default sampling params

Update generation_config with default sampling parameters

Rename classes to NemotronH_Nano_Omni_Reasoning_V3 for vLLM compatibility

Upload folder using huggingface_hub

initial commit