File size: 2,913 Bytes
c5f5bdd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 |
---
license: llama3.2
license_link: https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct/blob/main/LICENSE.txt
library: llama.cpp
library_link: https://github.com/ggerganov/llama.cpp
base_model:
- meta-llama/Llama-3.2-1B-Instruct
language:
- en
- de
- fr
- it
- pt
- hi
- es
- th
pipeline_tag: text-generation
tags:
- nlp
- code
- gguf
---
## LLaMA 3.2 1B Instruct
LLaMA 3.2 3B Instruct is a multilingual instruction-tuned language model with 3.21 billion parameters. Designed for diverse multilingual dialogue and summarization tasks, it offers effective performance on a range of NLP benchmarks.
### Model Information
- **Name**: LLaMA 3.2 3B Instruct
- **Parameter Size**: 3B (3.21B)
- **Model Family**: LLaMA 3.2
- **Architecture**: Auto-regressive Transformer with Grouped-Query Attention (GQA)
- **Purpose**: Multilingual dialogue generation, text generation, and summarization.
- **Training Data**: A mix of publicly available multilingual data, covering up to 9T tokens.
- **Supported Languages**: English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai.
- **Release Date**: September 25, 2024
- **Context Length**: 128k tokens
- **Knowledge Cutoff**: December 2023
### Quantized Model Files
- **Available Formats**:
- **ggml-model-q8_0.gguf**: 8-bit quantization for resource efficiency and good performance.
- **ggml-model-f16.gguf**: Half-precision (16-bit) floating-point format for enhanced precision.
- **Quantization Library**: llama.cpp
- **Use Cases**: Multilingual dialogue, summarization, and text generation.
### Core Library
LLaMA 3.2 1B Instruct can be deployed using `llama.cpp` or `transformers`, with a focus on streamlined integration into the Hugging Face ecosystem.
- **Primary Framework**: `llama.cpp`
- **Alternate Frameworks**:
- `transformers` for Hugging Face model support.
- `vLLM` for optimized inference and low-latency deployments.
**Library and Model Links**:
- **Model Base**: [meta-llama/Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct)
- **Models**: [meta-llama/llama-stack](https://github.com/meta-llama/llama-stack)
- **Inference Support**: [meta-llama/llama](https://github.com/meta-llama/llama)
- **Quantization**: [ggerganov/llama.cpp](https://github.com/ggerganov/llama.cpp)
### Safety and Responsible Use
LLaMA 3.2 3B has been designed with safety in mind but may produce biased, harmful, or unpredictable outputs, especially for less-covered languages or specific prompts.
- **Testing and Risk Assessment**: Initial testing has primarily focused on English; coverage for other languages is ongoing.
- **Limitations**: LLaMA 3.2 may not fully adhere to user instructions or safety guidelines, and may exhibit unexpected behaviors.
- **Responsible Use Guidelines**: Refer to the [Responsible Use Guide](https://ai.meta.com/llama/responsible-use-guide/) for more details.
|