luckeciano/Qwen-2.5-7B-GRPO-LR-3e-5-Adam-HessianMaskToken-2e-2-Symmetric-v2_5800

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 13, 2025Architecture:Transformer Featherless Exclusive Cold

The luckeciano/Qwen-2.5-7B-GRPO-LR-3e-5-Adam-HessianMaskToken-2e-2-Symmetric-v2_5800 is a 7.6 billion parameter language model, fine-tuned from Qwen/Qwen2.5-Math-7B. It was trained using the GRPO method on the DigitalLearningGmbH/MATH-lighteval dataset, specializing it for mathematical reasoning tasks. This model is designed to enhance performance in complex mathematical problem-solving, leveraging its specialized training for improved accuracy and understanding in this domain.

Loading preview...

Overview

This model, luckeciano/Qwen-2.5-7B-GRPO-LR-3e-5-Adam-HessianMaskToken-2e-2-Symmetric-v2_5800, is a specialized 7.6 billion parameter language model. It is a fine-tuned variant of the Qwen/Qwen2.5-Math-7B base model, specifically optimized for mathematical reasoning. The training utilized the GRPO (Generalized Reinforcement Learning for Policy Optimization) method, as introduced in the DeepSeekMath research, and was conducted on the DigitalLearningGmbH/MATH-lighteval dataset.

Key Capabilities

  • Enhanced Mathematical Reasoning: Specialized fine-tuning with the GRPO method on a math-focused dataset significantly improves its ability to understand and solve complex mathematical problems.
  • Foundation on Qwen 2.5: Benefits from the robust architecture and general language understanding capabilities of the Qwen 2.5 series.
  • Efficient Training Method: Leverages the TRL library for efficient and effective fine-tuning, incorporating advanced reinforcement learning techniques.

Good For

  • Mathematical Problem Solving: Ideal for applications requiring high accuracy in arithmetic, algebra, geometry, and other advanced mathematical domains.
  • Research in Mathematical AI: Useful for researchers exploring the application of reinforcement learning and specialized fine-tuning techniques to improve mathematical capabilities in large language models.
  • Educational Tools: Can be integrated into platforms designed to assist with or generate solutions for mathematical challenges.

Finetunes
1 models