Overview

hemanth-kj/llama-2-7B is a 7 billion parameter model from Meta's Llama 2 family of large language models. It is a pretrained, auto-regressive language model built on an optimized transformer architecture, designed to generate text. The Llama 2 models were trained on a new mix of publicly available online data, totaling 2 trillion tokens, with a data cutoff of September 2022.

Key Capabilities

• Text Generation: Capable of generating human-like text based on input prompts.
• Foundation Model: Serves as a base model that can be adapted for various natural language generation tasks.
• Optimized Architecture: Utilizes an optimized transformer architecture for efficient performance.

Intended Use Cases

• Commercial and Research: Suitable for both commercial applications and academic research in English.
• Natural Language Generation: Can be fine-tuned or adapted for a wide range of NLG tasks.

Performance Highlights

Compared to Llama 1 7B, Llama 2 7B shows improvements across several academic benchmarks, including:

• Code: Improved from 14.1 to 16.8.
• Commonsense Reasoning: Improved from 60.8 to 63.9.
• Math: Significantly improved from 6.95 to 14.6.
• MMLU: Improved from 35.1 to 45.3.

Limitations

• Language: Primarily intended for use in English.
• Static Model: Trained on an offline dataset, meaning its knowledge cutoff is September 2022 for pretraining data.
• Safety: As with all LLMs, it may produce inaccurate, biased, or objectionable responses, requiring developers to perform safety testing and tuning.