Llama 3.1 Nemotron 51BOpen Weight

Released

Llama 3.1 Nemotron 51B is an AI model released by NVIDIA on Monday, Sep 23 2024, 33 days after Mistral NeMo Minitron 8B. It is an open-weight model — the trained weights are available to download and run. It is a 51B parameterThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. model with a 128k token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover ARC-Challenge, HellaSwag, WinoGrande, MMLU, and GSM8K.

Benchmarks

ARC-ChallengeReasoning — multiple-choice science questions from the AI2 Reasoning Challenge; unrelated to ARC-AGI
69.2%
#8 of 14
HellaSwagReasoning — commonsense completion of short scenarios
85.58%
#10 of 22
WinoGrandeReasoning — resolving ambiguous pronouns in context
84.53%
#4 of 13
MMLUGeneral knowledge — A 57-subject multiple-choice exam — history, law, medicine, maths — that was the standard measure of how much a model knows from 2020 until roughly 2024, when frontier scores crowded into the high 80s and labs moved on to harder tests. The scores here were published years apart under different testing setups, so read them as a historical record rather than a like-for-like ranking. Higher is better.
80.2%
#43 of 81
GSM8KGrade-school math — Grade-school maths word problems that take a few steps of arithmetic to work through. It separated the models of 2022 and 2023 sharply, then saturated. One caveat on the historical numbers: OpenAI included part of the GSM8K training set in GPT-4's pre-training mix, so GPT-4's score is not a clean few-shot result. Higher is better.
91.43%
#18 of 56

Compare Llama 3.1 Nemotron 51B with

Llama 3.1 Nemotron 51B

About

Llama 3.1 Nemotron 51B was released in September 2024 as a compressed derivative of Meta’s Llama 3.1 70B Instruct. NVIDIA used neural architecture search and block distillation to target single-GPU inference.

The 51B Instruct checkpoint shipped with an NVIDIA NIM deployment and downloadable weights. Its model card reported 80.2% on five-shot MMLU and 91.43% on five-shot GSM8K.

Frequently asked questions

Llama 3.1 Nemotron 51B was released by NVIDIA on Monday, Sep 23 2024.

All NVIDIA releases

20 tracked