aayanmishra-ml/Athena-1-3B

Athena-1 3B Overview

Athena-1 3B is a compact yet powerful instruction-following large language model developed by aayanmishra-ml, fine-tuned from the Qwen2.5-3B-Instruct base. With just 3.09 billion parameters, it is designed for efficiency and high-quality text generation in resource-constrained environments. The model supports a substantial 32,768 token context length, allowing it to process moderately long documents and conversations, and can generate up to 8K tokens of output.

Key Capabilities

• Lightweight and Efficient: Offers strong performance with reduced computational demands due to its compact size.
• Instruction Following: Precisely adheres to user prompts for reliable output generation.
• Coding and Mathematics: Demonstrates proficiency in solving coding challenges and handling mathematical tasks.
• Long-Context Understanding: Processes and understands information across a 32,768 token context window.
• Multilingual Support: Capable of operating in over 29 languages, including English, Chinese, French, Spanish, Japanese, and Korean.
• Structured Data Processing: Interprets and generates structured formats like tables and JSON, making it suitable for data-centric applications.

Ideal Use Cases

• Conversational AI: Building fast, responsive, and lightweight chatbots.
• Code Generation: Generating, debugging, or explaining code snippets.
• Mathematical Problem Solving: Assisting with calculations and logical reasoning.
• Document Processing: Summarizing and analyzing moderately sized documents.
• Multilingual Applications: Supporting global use cases with diverse language requirements.
• Structured Data Tasks: Processing and generating structured data outputs, such as JSON.