Llama-3_1-Nemotron-Ultra-253B-CPT-v1

Home/Models/nvidia/Llama-3_1-Nemotron-Ultra-253B-CPT-v1

nvidia/Llama-3_1-Nemotron-Ultra-253B-CPT-v1

kind: original
quant: bf16 (16-bit)
family: llama
modality: text-generation
root model: nvidia/Llama-3_1-Nemotron-Ultra-253B-CPT-v1
provider: nvidia
license: other
language: en
context: 128K
library: transformers
vRAM (GB): —
size on disk: —
downloads: 83
likes: 6
created: —
last modified: 2025-07-06T15:51:59+00:00 (2025)
huggingface: https://huggingface.co/nvidia/Llama-3_1-Nemotron-Ultra-253B-CPT-v1
tags: transformerssafetensorsnemotron-nastext-generationnvidiallama-3pytorchconversationalcustom_codeenarxiv:2503.18908arxiv:2411.19146license:otherregion:us

benchmark	category	score	context	conf	evidence
HumanEval	coding	88.41	HumanEval	medium	`\| HumanEval \| pass@1 \| 88.41 \|`
MMLU	knowledge	88.09	MMLU	medium	`\| MMLU \| macro \| 88.09 \|`
RULER	long-context	83.21	RULER	medium	`\| RULER \| 128K \| 83.21 \|`
MATH-500	math	80.4	MATH500	medium	`\| MATH500 \| micro \| 80.4 \|`

none scraped.