kmseong/RSN-GSM8K-SFT-Model
Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kPublished:Dec 13, 2025License:llama3.1Architecture:Transformer Warm

The kmseong/RSN-GSM8K-SFT-Model is an 8 billion parameter Llama 3.1 Instruct model, fine-tuned by kmseong using LoRA on the GSM8K dataset. This model is specifically optimized for mathematical reasoning tasks, demonstrating a 55.00% accuracy on the GSM8K test set. It is designed to enhance problem-solving capabilities for arithmetic and word problems.

Loading preview...

Model Overview

The kmseong/RSN-GSM8K-SFT-Model is an 8 billion parameter language model based on meta-llama/Llama-3.1-8B-Instruct. It has been fine-tuned by kmseong using Low-Rank Adaptation (LoRA) on the GSM8K dataset to significantly improve its mathematical reasoning abilities.

Key Capabilities

  • Enhanced Mathematical Reasoning: Specifically trained on 500 samples from the GSM8K dataset, focusing on arithmetic and word problems.
  • Llama 3.1 Base: Leverages the strong foundational capabilities of the Llama 3.1 Instruct architecture.
  • LoRA Fine-tuning: Utilizes an efficient LoRA configuration (rank 8, alpha 16) targeting key attention and feed-forward modules.
  • Performance: Achieves 55.00% accuracy on the GSM8K test set, indicating its specialized proficiency in mathematical problem-solving.

Good For

  • Mathematical Problem Solving: Ideal for applications requiring step-by-step mathematical reasoning and accurate numerical answers.
  • Educational Tools: Can be integrated into systems designed to help users solve or understand math problems.
  • Research in Mathematical LLMs: Provides a fine-tuned base for further experimentation and development in specialized reasoning tasks.

Limitations

This model is fine-tuned specifically for mathematical reasoning. Its performance on general-purpose language tasks or other domains may vary and should be evaluated for specific use cases.