ertghiu256/Qwen3-4b-2507-Thinking-math-and-code

TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Oct 4, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

The ertghiu256/Qwen3-4b-2507-Thinking-math-and-code is a 4 billion parameter Qwen 3 model developed by ertghiu256, fine-tuned for enhanced reasoning, mathematical tasks, and code generation. This model leverages Unsloth for faster training and is specifically optimized for handling complex math problems and code samples. It is designed to excel in applications requiring strong analytical and programming capabilities, building upon its 32768 token context length.

Loading preview...

Model Overview

The ertghiu256/Qwen3-4b-2507-Thinking-math-and-code is a 4 billion parameter Qwen 3 language model, developed by ertghiu256. It has been specifically fine-tuned from unsloth/qwen3-4b-thinking-2507-unsloth-bnb-4bit to improve its performance in mathematical reasoning and code-related tasks.

Key Capabilities

  • Enhanced Mathematical Reasoning: Optimized for solving complex mathematical problems.
  • Code Generation and Understanding: Proficient in handling and generating code samples.
  • Efficient Training: Fine-tuned using Unsloth and Huggingface's TRL library, enabling 2x faster training.
  • Context Length: Supports a substantial context window of 32768 tokens.

Training Details

The model was trained on the ertghiu256/MathReasoning-with-code-samples dataset over 150 steps, utilizing a learning rate of 6e-5. This specialized training regimen contributes to its focused capabilities in analytical and programming domains.

Good For

  • Applications requiring strong mathematical problem-solving.
  • Code generation, completion, and understanding tasks.
  • Scenarios where efficient reasoning and analytical capabilities are crucial.