mlx-community/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-bf16

Same weights, converted vs nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16. Same weights stored in another dtype (BF16/F32 → BF16), plus 49 new tensors.

Tensors changed: Norms 0/543, Embeddings 0/4, other 0/194, MLP 0/270, Attention 0/320, lm_head 0/1

2 derivatives checked