Qwen2.5-14B-Instruct-1M-GRPO-Reasoning

notbdq

Parameters

14.8B

Downloads

—

Likes

—

Avg score

41.6

6 benchmarks

Weights

Open

Released

—

About

Qwen2.5-14B-Instruct-1M-GRPO-Reasoning is a 14.8B-parameter open-weights model in the Qwen2 family from notbdq. Benchmarked on 6 evals, averaging 41.6.

Benchmark scores · 6

avg 41.6 — higher is better

IFEvalOpen LLM LB · 2026-05

84.1

MATH Lvl 5Open LLM LB · 2026-05

53.0

BIG-Bench HardOpen LLM LB · 2026-05

45.7

MMLU-ProOpen LLM LB · 2026-05

42.8

GPQA DiamondOpen LLM LB · 2026-05

12.4

MuSROpen LLM LB · 2026-05

11.3

Details

Publishernotbdq

FamilyQwen2

Parameters14.8B

License

WeightsOpen

Released—

History

full ledger →

Capability over time

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