Llama-3.1-8B-GRPO-Instruct

ymcki

Parameters

Downloads

—

Likes

—

Avg score

28.2

6 benchmarks

Weights

Open

Released

—

About

Llama-3.1-8B-GRPO-Instruct is a 8B-parameter open-weights model in the Llama family from ymcki. Benchmarked on 6 evals, averaging 28.2.

llama3.1

Benchmark scores · 6

avg 28.2 — higher is better

IFEvalOpen LLM LB · 2026-05

74.5

MMLU-ProOpen LLM LB · 2026-05

30.4

BIG-Bench HardOpen LLM LB · 2026-05

30.4

MATH Lvl 5Open LLM LB · 2026-05

20.2

MuSROpen LLM LB · 2026-05

7.6

GPQA DiamondOpen LLM LB · 2026-05

5.9

Details

Publisherymcki

FamilyLlama

Parameters8B

Licensellama3.1

WeightsOpen

Released—

History

Capability over time

Append-only ledger — every observed change to this model's numbers.