L

Llama 3.2 3B Instruct

Developed by meta-llama
Llama 3.2 is a multilingual large language model series developed by Meta, including 1B and 3B scale pre-trained and instruction-tuned generative models, optimized for multilingual conversation scenarios.
Downloads 1.6M
Release Time : 9/18/2024

Model Overview

Llama 3.2 is an instruction-tuned pure text generation model suitable for multilingual conversations, agent retrieval, and summarization tasks, with excellent performance in multiple benchmarks.

Model Features

Multilingual support
Officially supports 8 languages, with a broader training language range, allowing developers to fine-tune for other languages.
Long-context processing
Supports 128k context length, suitable for handling long documents and complex conversation scenarios.
Efficient quantization
Provides 4-bit group quantization and SpinQuant technology, optimizing mobile deployment and inference speed.
Human preference alignment
Achieves alignment with human preferences through supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF).

Model Capabilities

Multilingual text generation
Code generation
Long document processing
Dialogue systems
Summarization generation
Information retrieval

Use Cases

Dialogue systems
Multilingual customer service chatbot
Build intelligent customer service systems supporting multiple languages, delivering natural and fluent conversation experiences.
Performs excellently in MMLU tests for languages like Spanish and Hindi.
Content generation
Multilingual content creation
Automatically generate marketing copy, news summaries, and other content in multiple languages.
Performs well in GSM8K math problems and ARC-C reasoning tests.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase