Nemotron-H-47B-Reasoning-128K-FP8

Home/Models/nvidia/Nemotron-H-47B-Reasoning-128K-FP8

nvidia/Nemotron-H-47B-Reasoning-128K-FP8

kind: fp8
quant: fp8 (8-bit)
family: nemotron
modality: text-generation
root model: nvidia/Nemotron-H-47B-Reasoning-128K
provider: nvidia
license: other
language: en
context: 128K
library: transformers
vRAM (GB): —
size on disk: —
downloads: 114
likes: 6
created: —
last modified: 2025-08-21T15:46:36+00:00 (2025)
huggingface: https://huggingface.co/nvidia/Nemotron-H-47B-Reasoning-128K-FP8
tags: transformerssafetensorsnvidiapytorchtext-generationconversationalenarxiv:2504.03624arxiv:2505.00949license:otherendpoints_compatibleregion:us

benchmark	category	score	context	conf	evidence
MATH-500	math	96.6	MATH500	medium	`\| MATH500 \| 96.60% \| 96.20% \| 96.80% \|`
MBPP	coding	90.2	MBPP	medium	`\| MBPP \| 90.20% \| 91.80% \| 91.80% \|`
IFEval	instruction-following	87.4	IFEVAL-Instruction	medium	`\| IFEVAL-Instruction \| 87.40% \| 89.69% \| 89.57% \|`
Chatbot Arena	chat	85.0	Arena Hard	medium	`\| Arena Hard \| 85.00% \| 85.00% \| - \|`
IFEval	instruction-following	81.8	IFEVAL-Prompt	medium	`\| IFEVAL-Prompt \| 81.80% \| 84.47% \| 84.47% \|`
BFCL (function calling)	agentic	72.68	BFCL	medium	`\| BFCL \| 72.68% \| 72.95% \| 73.21% \|`
GPQA	reasoning	65.1	GPQA	medium	`\| GPQA \| 65.10% \| 65.66% \| 64.65% \|`
AIME 2025	math	53.0	AIME25	medium	`\| AIME25 \| 53.00% \| 54.17% \| 54.17% \|`

none scraped.