Model Overview

Neelectric/Llama-3.1-8B-Instruct_SDFT_mathv00.07 is an 8 billion parameter instruction-tuned model, building upon the strong foundation of Meta's Llama-3.1-8B-Instruct. Developed by Neelectric, this model has been specifically fine-tuned using the Neelectric/OpenR1-Math-220k_all_SDFT_nr dataset, making it particularly adept at mathematical reasoning.

Key Training Details

• Base Model: Meta's Llama-3.1-8B-Instruct
• Fine-tuning Method: Utilizes SDFT (Self-Training with On-Policy Self-Distillation), a technique designed for language model alignment, as detailed in the paper "Self-Training with On-Policy Self-Distillation for Language Model Alignment".
• Training Framework: Trained using the TRL library.
• Context Length: Supports a substantial context window of 32768 tokens.

Use Cases

This model is primarily suited for applications requiring:

• Mathematical Problem Solving: Excels in tasks that involve numerical reasoning and mathematical operations.
• Instruction Following: Benefits from its instruction-tuned base, allowing for precise responses to user prompts.
• Research and Development: Ideal for researchers exploring self-training and self-distillation methods in LLMs, particularly in the domain of mathematics.