Overview

NousResearch/Llama-2-7b-hf is a 7 billion parameter pretrained model from Meta's Llama 2 family of large language models. This specific model is provided in the Hugging Face Transformers format. The Llama 2 collection includes both pretrained and fine-tuned variations, with the fine-tuned Llama-2-Chat models optimized for dialogue use cases.

Key Capabilities

• Architecture: Utilizes an optimized transformer architecture for auto-regressive text generation.
• Scale: This 7B parameter model is part of a family that also includes 13B and 70B parameter versions.
• Training Data: Pretrained on 2 trillion tokens of a new mix of publicly available online data, with a data cutoff of September 2022.
• Context Length: Supports a context length of 4096 tokens.
• Performance: Demonstrates improved performance over Llama 1 models across various academic benchmarks, including Code, Commonsense Reasoning, World Knowledge, Reading Comprehension, Math, MMLU, and BBH.

Intended Use Cases

• Research and Commercial Use: Designed for a broad range of research and commercial applications in English.
• Natural Language Generation: As a pretrained model, it can be adapted for diverse natural language generation tasks.
• Base Model: Serves as a foundational model for further fine-tuning or specialized applications.