phamhai/Llama-3.2-1B-Instruct-Frog
Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:1BQuant:BF16Ctx Length:32kPublished:Oct 22, 2024License:llama3.2Architecture:Transformer0.0K Warm

The phamhai/Llama-3.2-1B-Instruct-Frog is a 1 billion parameter instruction-tuned causal language model, based on Meta's Llama-3.2-1B-Instruct architecture, with a 131K context length. It is specifically optimized for Vietnamese language tasks, particularly for Retrieval-Augmented Generation (RAG) applications. This model is designed for fast inference and efficient deployment on edge devices.

Loading preview...

Overview

phamhai/Llama-3.2-1B-Instruct-Frog is a 1 billion parameter instruction-tuned model, built upon Meta's Llama-3.2-1B-Instruct. It features an extended context length of 131K tokens. The primary focus of this model is to provide enhanced support for Vietnamese language tasks, especially within Retrieval-Augmented Generation (RAG) frameworks, where the base Llama-3.2 models showed limitations.

Key Capabilities

  • Vietnamese RAG Optimization: Specifically trained to improve performance in Vietnamese RAG scenarios.
  • Efficient Deployment: Optimized for fast inference, making it suitable for deployment on on-premise and edge devices such as laptops, smartphones, NVIDIA Jetson Xavier, and Raspberry Pi.
  • Instruction Following: Capable of handling various instruction-based tasks, including Question Answering, Summarization, Function Calling, and Question Rewriting, as demonstrated by provided examples.

Limitations

  • The 1B parameter version may exhibit poorer prompt understanding and lower accuracy in certain tasks like summarization and entity extraction in Function Calling compared to larger models.

Good For

  • Developers building Vietnamese-centric RAG applications.
  • Projects requiring a lightweight LLM for edge device deployment.
  • Use cases where fast inference and Vietnamese language proficiency are critical.