A math-specific LLM

Aug 9， 2024

The Qwen Team recently introduced Qwen2-Math, a new series of math-focused language models. These models, ranging from 1.5B to 72B parameters, demonstrate superior performance on various math benchmarks, surpassing even the most advanced closed-source models like GPT-4. Qwen2-Math models are trained on extensive, high-quality mathematical datasets, and are fine-tuned for complex problem-solving. Evaluations on both English and Chinese math benchmarks reveal their exceptional capability, making them valuable tools for tackling advanced mathematical problems.

For more details, visit the original article.