Model CoverageLarge Language ModelsNVIDIANVIDIA-Nemotron-3-Nano-4B-BF16

NVIDIA-Nemotron-3-Nano-4B-BF16

View as Markdown

NVIDIA Nemotron-H is a hybrid Mamba-2 and transformer architecture that interleaves selective state-space layers with standard attention layers for improved efficiency on long sequences.

Set up NeMo AutoModel with the latest container or follow the installation instructions.

Fine-Tune NVIDIA-Nemotron-3-Nano-4B-BF16

From the repository root, run:

uv run automodel examples/llm_finetune/nemotron/nemotron_nano_4b_squad.yaml \
  --nproc-per-node 8

Choose a Workflow

GoalStart Here
Supervised fine-tuning (SFT) - Nemotron-3-Nano 4B on SQuADUse nemotron_nano_4b_squad.yaml.
Low-rank adaptation (LoRA) - Nemotron-3-Nano 4B on SQuADUse nemotron_nano_4b_squad_peft.yaml.

Model Reference

Model Architecture

PropertyValue
TaskText Generation
ArchitectureNemotronHForCausalLM
Parameters4B
Hugging Face Organizationnvidia

Available Models

ModelHF ID
Nemotron-3-Nano 4Bnvidia/NVIDIA-Nemotron-3-Nano-4B-BF16