Xmodel-2 Technical Report

TL;DR

Xmodel-2 is a large language model with 1.2 billion parameters, optimized for reasoning tasks, employing a unified hyperparameter approach and WSD scheduler, achieving state-of-the-art results efficiently.

Contribution

Introduces Xmodel-2, a reasoning-focused language model with a unified hyperparameter design and effective training strategies for improved performance.

Findings

Achieves state-of-the-art reasoning performance

Maintains low training costs

Demonstrates effective transfer of configurations across model scales

Abstract

Xmodel-2 is a 1.2-billion-parameter large language model designed specifically for reasoning tasks. Its architecture enables different model scales to share a unified set of hyperparameters, allowing for extensive experimentation on smaller models and seamless transfer of optimal configurations to larger models. To maximize training efficiency and stability, Xmodel-2 employs the WSD learning rate scheduler from MiniCPM. Pretrained on 1.5 trillion tokens from diverse sources, Xmodel-2 achieves state-of-the-art performance in complex reasoning and agent-based tasks, while maintaining low training costs. These results highlight the potential of efficient model design and training strategies in advancing reasoning capabilities. Model checkpoints and code are publicly available on GitHub at https://github.com/XiaoduoAILab/Xmodel-2