MDL
Qwen2.5-3B-Instruct
Qwen
Parameters
3B
Downloads
7.1M
on HF
Likes
532
Avg score
27.2
6 benchmarks
Weights
Open
Released
Sep 17, 2024
About
Qwen2.5-3B-Instruct is a 3B-parameter open-weights model in the Qwen2 family from Qwen. Benchmarked on 6 evals, averaging 27.2.
othertext-generation
Benchmark scores · 6
avg 27.2 — higher is betterThe research behind this model
all papers mentioning it →Qwen2 Technical ReportRoPE-Aware Bit Allocation for KV-Cache Quantization▲ 8 on HF · Jun 23, 2026StepOPSD: Step-Aware Online Preference Distillation for Agent Reinforcement LearningMay 26, 2026Distilling Game Code World Model Generation into Lightweight Large Language ModelsMay 23, 2026Mechanistic origins of catastrophic forgetting: why RL preserves circuits better than SFT?May 21, 2026DASH: Fast Differentiable Architecture Search for Hybrid Attention in Minutes on a Single GPUMay 20, 2026
Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.
History
full ledger →Capability over time
Append-only ledger — every observed change to this model's numbers.