NVIDIA-Nemotron-Nano-9B-v2-FP8

Home/Models/nvidia/NVIDIA-Nemotron-Nano-9B-v2-FP8

nvidia/NVIDIA-Nemotron-Nano-9B-v2-FP8

kind: fp8
quant: fp8 (8-bit)
family: nemotron
modality: text-generation
root model: nvidia/NVIDIA-Nemotron-Nano-9B-v2
provider: nvidia
license: other
language: en, es, fr, de, it, ja
context: 128K
library: transformers
vRAM (GB): —
size on disk: —
downloads: 128,501
likes: 9
created: —
last modified: 2026-01-08T19:23:12+00:00 (2026)
huggingface: https://huggingface.co/nvidia/NVIDIA-Nemotron-Nano-9B-v2-FP8
tags: transformerssafetensorsnemotron_htext-generationnvidiapytorchconversationalcustom_codeenesfrdeitjadataset:nvidia/Nemotron-Post-Training-Dataset-v1dataset:nvidia/Nemotron-Post-Training-Dataset-v2dataset:nvidia/Nemotron-Pretraining-Dataset-sampledataset:nvidia/Nemotron-CC-v2dataset:nvidia/Nemotron-CC-Math-v1dataset:nvidia/Nemotron-Pretraining-SFT-v1arxiv:2504.03624arxiv:2508.14444arxiv:2412.02595base_model:nvidia/NVIDIA-Nemotron-Nano-9B-v2

benchmark	category	score	context	conf	evidence
MATH-500	math	96.3	MATH500	medium	`\| MATH500 \| 96.3% \| 97.8% \| 97.5% \|`
IFEval	instruction-following	89.4	IFEval (Instruction Strict)	medium	`\| IFEval (Instruction Strict) \| 89.4% \| 90.3% \| 90.2% \|`
RULER	long-context	74.1	RULER (128K)	medium	`\| RULER (128K) \| 74.1% \| 78.9% \| 76.1% \|`
AIME 2025	math	69.3	AIME25	medium	`\| AIME25 \| 69.3% \| 72.1% \| 71.3% \|`
BFCL (function calling)	agentic	66.3	BFCL v3	medium	`\| BFCL v3 \| 66.3% \| 66.9% \| 65.7% \|`
GPQA	reasoning	59.6	GPQA	medium	`\| GPQA \| 59.6% \| 64.0% \| 60.6% \|`

none scraped.