Model Overview

MaziyarPanahi/Llama-3-8B-Instruct-v0.10 is an instruction-tuned large language model built upon the Llama 3 architecture, featuring 8 billion parameters. This version is an enhancement over its predecessor, Llama-3-8B-Instruct-v0.9, developed by MaziyarPanahi.

Key Capabilities & Features

• Instruction Following: Optimized for understanding and executing a wide range of user instructions.
• Conversational AI: Designed for interactive chat applications, leveraging the ChatML prompt template for structured dialogues.
• Context Length: Supports an 8192-token context window, allowing for more extensive conversations and complex prompts.
• Quantized GGUF: Available in quantized GGUF formats for efficient deployment and inference on various hardware.

Performance Highlights

Evaluated on the Open LLM Leaderboard, the model achieved an average score of 26.66. Notable scores include 76.67 on IFEval (0-Shot) and 31.80 on MMLU-PRO (5-shot), indicating its proficiency in instruction understanding and general knowledge tasks.

Usage

This model can be readily integrated using the Hugging Face transformers library, with provided Python code examples demonstrating how to set up the pipeline for text generation and apply the ChatML prompt template.