Corianas/tiny-llama-miniguanaco-1.5T: Question Answering Model

This model, developed by Corianas, is a 1.1 billion parameter variant of TinyLlama, specifically fine-tuned for question-answering tasks. It is designed to process a prompt (question) and generate a direct answer, concluding the response with a <END> token on a new line.

Key Capabilities

• Direct Question Answering: Optimized to provide concise answers to user queries.
• Compact Size: With 1.1 billion parameters, it offers a relatively small footprint for deployment.
• Structured Output: Responses are formatted with a clear question-then-answer structure, ending with <END>.

Usage and Formatting

The model expects a simple input format: "prompt"\n"completion"\n<END>. When generating text, the prompt should be followed by a newline to initiate the answer. The output will contain the answer, terminated by <END> on a new line.

Good For

• Applications requiring straightforward, factual answers to questions.
• Scenarios where a compact, efficient language model for question answering is beneficial.
• Integration into systems that can parse the <END> token for response termination.