Mistral NeMo Minitron 8BOpen Weight

Released

Mistral NeMo Minitron 8B is an AI model released by NVIDIA on Wednesday, Aug 21 2024, 68 days after Nemotron-4 340B. It is an open-weight model — the trained weights are available to download and run. It is a 8B 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 8k 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 HumanEval, ARC-Challenge, HellaSwag, WinoGrande, MMLU, and GSM8K.

Benchmarks

Coding

HumanEvalFunction synthesis — 164 small Python problems: the AI is given a function's description and has to write the working function. This was the coding benchmark of the GPT-3.5 and GPT-4 era, before the field moved to fixing real bugs in real repositories. Higher is better.
36.22%
#77 of 88

Reasoning & science

ARC-ChallengeReasoning — multiple-choice science questions from the AI2 Reasoning Challenge; unrelated to ARC-AGI
64.42%
#9 of 14
HellaSwagReasoning — commonsense completion of short scenarios
83.03%
#16 of 22
WinoGrandeReasoning — resolving ambiguous pronouns in context
80.35%
#6 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.
69.51%
#63 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.
58.45%
#47 of 56

Compare Mistral NeMo Minitron 8B with

Mistral NeMo Minitron 8B

About

NVIDIA released Mistral NeMo Minitron 8B on August 21, 2024. Mistral AI and NVIDIA jointly developed the original 12B Mistral NeMo. NVIDIA pruned that model to 8B parameters and trained the smaller model using knowledge distillation.

The 8B base checkpoint targeted workloads that could run on a workstation. NVIDIA tested it on language understanding, math and code generation and reported 69.51% on MMLU.

Frequently asked questions

Mistral NeMo Minitron 8B was released by NVIDIA on Wednesday, Aug 21 2024.

All NVIDIA releases

20 tracked