unsloth/NVIDIA-Nemotron-3-Ultra-550B-A55B-GGUF

Same weights, converted vs nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16. Same weights stored in another dtype (BF16/F32 → BF16/F32), plus 192 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/109, lm_head 0/1

1 derivatives checked

• meshllm/NVIDIA-Nemotron-3-Ultra-550B-A55B-UD-Q4_K_XL-layers: Quantization (fewer layers)

On the Hub · Downloads on Model Pulse