burgasdotpro/bgGPT-Qwen2.5-Math-7B-Inst

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kLicense:apache-2.0Architecture:Transformer0.0K Open Weights Warm

The burgasdotpro/bgGPT-Qwen2.5-Math-7B-Inst is a 7.6 billion parameter instruction-tuned causal language model developed by burgasdotpro, fine-tuned from unsloth/qwen2.5-math-7b-instruct-bnb-4bit. This model is specifically optimized for mathematical reasoning and problem-solving in Bulgarian, trained on 7.5K lines of mathematical data extracted and optimized from the mathqa-bgeval dataset. It excels at processing and generating solutions for mathematical questions, making it suitable for applications requiring numerical and logical problem-solving capabilities.

Loading preview...

bgGPT-Qwen2.5-Math-7B-Inst Overview

The burgasdotpro/bgGPT-Qwen2.5-Math-7B-Inst is a 7.6 billion parameter instruction-tuned language model developed by burgasdotpro. It is fine-tuned from the unsloth/qwen2.5-math-7b-instruct-bnb-4bit base model, with a specific focus on mathematical tasks.

Key Capabilities

  • Specialized Mathematical Reasoning: The model has been trained on 7.5K lines of mathematical data, optimized from the mathqa-bgeval dataset provided by INSAIT. This specialization allows it to accurately solve and explain mathematical problems.
  • Bulgarian Language Support: Its training data and examples demonstrate proficiency in understanding and generating responses for mathematical queries in Bulgarian.
  • Efficient Fine-tuning: The model was fine-tuned using Unsloth and the Hugging Face TRL library, enabling faster training times.

Good For

  • Mathematical Problem Solving: Ideal for applications requiring the solution of arithmetic and algebraic problems.
  • Educational Tools: Can be integrated into platforms for teaching or assisting with mathematics in Bulgarian.
  • Bulgarian NLP with Math Focus: Suitable for tasks that combine natural language processing in Bulgarian with a strong mathematical component.

Usage Template

The model follows a specific instruction template for optimal performance:

### Instruction:
{}

### Input:
{}

### Response:
{}