LLM ReferenceLLM Reference

Granite 3.1 2B Instruct

granite-3.1-2b-instruct

Open Source

About

IBM Granite 3.1 2B Instruct (2.5B actual parameters) is a compact instruction-tuned model with 128K context window. Dense decoder-only transformer with 40 layers, 2048 embedding, GQA. Trained on 12 trillion tokens. Supports 12 languages, code, RAG, and long-context tasks. Used as the LLM backbone for Granite Vision 3.2 2B. Apache 2.0.

Granite 3.1 2B Instruct has a 128K-token context window.

Capabilities

VisionMultimodalReasoningFunction CallingTool UseStructured OutputsCode Execution

Rankings

Specifications

FamilyGranite 3
Released2024-12-18
Parameters2B
Context128K
ArchitectureDense decoder-only transformer: 40 layers, 2048 embed, GQA 32/8 heads, RoPE, SwiGLU

Created by

Creating reliable and adaptable AI solutions

Armonk, New York, United States
Founded 1945
Website