Nemotron-H-8B-Base-8K

Home/Models/nvidia/Nemotron-H-8B-Base-8K

nvidia/Nemotron-H-8B-Base-8K

kind: original
quant: bf16 (16-bit)
family: nemotron
modality: text-generation
root model: nvidia/Nemotron-H-8B-Base-8K
provider: nvidia
license: other
language: en, de, es, fr, it, ko, pt, ru, jp, zh
context: 8K
library: transformers
vRAM (GB): —
size on disk: —
downloads: 86,116
likes: 59
created: —
last modified: 2025-08-21T15:45:04+00:00 (2025)
huggingface: https://huggingface.co/nvidia/Nemotron-H-8B-Base-8K
tags: transformerssafetensorsnvidiapytorchnemotron-htext-generationconversationalendeesfritkoptrujpzharxiv:2504.03624license:otherendpoints_compatibleregion:us

benchmark	category	score	context	conf	evidence
GSM8K	math	87.11	GSM8K 8-shot CoT \| MATH 4-shot CoT \| MATH-Lvl 5 4-shot CoT \| MATH-500 4-shot CoT	low	`\| 87.11 \| 46.52 \| 22.93 \| 44.43 \|`
WinoGrande	knowledge	80.51		low	`\| 88.74 \| 83.23 \| 80.51 \| 78.71 \|`
CommonsenseQA	knowledge	78.71		low	`\| 88.74 \| 83.23 \| 80.51 \| 78.71 \|`
MMLU	knowledge	72.77	MMLU-Pro 5-shot CoT	low	`\| 44.01 \| 72.77 \|`
MBPP	coding	65.37		low	`\| 65.37 \| 59.52 \| 58.54 \| 55.49 \|`
HumanEval	coding	58.54		low	`\| 65.37 \| 59.52 \| 58.54 \| 55.49 \|`
MATH	math	46.52	GSM8K 8-shot CoT \| MATH 4-shot CoT \| MATH-Lvl 5 4-shot CoT \| MATH-500 4-shot CoT	low	`\| 87.11 \| 46.52 \| 22.93 \| 44.43 \|`
MATH-500	math	44.43	GSM8K 8-shot CoT \| MATH 4-shot CoT \| MATH-Lvl 5 4-shot CoT \| MATH-500 4-shot CoT	low	`\| 87.11 \| 46.52 \| 22.93 \| 44.43 \|`
MMLU-Pro	knowledge	44.01	MMLU-Pro 5-shot CoT	low	`\| 44.01 \| 72.77 \|`
ARC-Challenge	knowledge	25.0		medium	`ARC Challenge 25`
HellaSwag	knowledge	10.0		medium	`Hellaswag 10`

none scraped.