Undi95/Meta-Llama-3-8B-Instruct-hf Overview
This model is an 8 billion parameter instruction-tuned variant from Meta's Llama 3 family, designed for generative text and code. It leverages an optimized transformer architecture and incorporates supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF) to align with human preferences for helpfulness and safety. The model was trained on over 15 trillion tokens of publicly available online data, with a knowledge cutoff of March 2023, and supports an 8k token context length.
Key Capabilities & Performance
- Dialogue Optimization: Specifically tuned for assistant-like chat applications, demonstrating strong performance on common industry benchmarks compared to other open-source chat models.
- Improved Safety & Helpfulness: Meta has focused on optimizing helpfulness and safety, significantly reducing false refusals to benign prompts compared to its predecessor, Llama 2.
- Benchmark Achievements: The instruction-tuned 8B model achieves 68.4 on MMLU, 62.2 on HumanEval, and 79.6 on GSM-8K, showcasing strong general reasoning, code generation, and mathematical capabilities.
- Scalability: Utilizes Grouped-Query Attention (GQA) for improved inference scalability.
Intended Use Cases
- Assistant-like Chat: Ideal for conversational AI and dialogue systems in commercial and research settings.
- Natural Language Generation: Suitable for various text generation tasks, particularly where instruction following and helpful responses are critical.
Meta emphasizes responsible AI development, providing resources like the Responsible Use Guide and tools such as Meta Llama Guard 2 and Code Shield to help developers implement safety best practices.