LLM ReferenceLLM Reference

Qwen2.5 3B Instruct

About

Instruction-tuned 3B variant optimized for following directions in educational and automation tools with compact memory footprint.

Capabilities

VisionMultimodalReasoningFunction CallingTool UseStructured OutputsCode Execution

Benchmark Scores(4)

BenchmarkScoreVersionSource
Google-Proof Q&A38.7diamondOpen LLM Leaderboard
HellaSwag86.210-shotOpen LLM Leaderboard
HumanEval55.2pass@1Open LLM Leaderboard
Massive Multitask Language Understanding75.25-shotOpen LLM Leaderboard

Rankings

Specifications

FamilyQwen2.5
Released2024-06-07
Parameters3.09B
Context128K
ArchitectureDecoder Only
Specializationgeneral
Trainingfinetuning
Fine-tuninginstruct

Created by

AI research institute of Alibaba Group.

Hangzhou, Zhejiang, China
Founded 2017
Website