Solar Open Technical Report

TL;DR

Solar Open is a large bilingual language model designed for underserved languages, utilizing a systematic data synthesis, curriculum optimization, and scalable reinforcement learning to achieve competitive performance.

Contribution

It introduces a comprehensive methodology combining data synthesis, curriculum learning, and RL optimization for building effective LLMs in low-resource languages.

Findings

Achieves competitive benchmarks in English and Korean.

Synthesizes 4.5T high-quality tokens for training.

Proposes SnapPO framework for efficient RL optimization.

Abstract

We introduce Solar Open, a 102B-parameter bilingual Mixture-of-Experts language model for underserved languages. Solar Open demonstrates a systematic methodology for building competitive LLMs by addressing three interconnected challenges. First, to train effectively despite data scarcity for underserved languages, we synthesize 4.5T tokens of high-quality, domain-specific, and RL-oriented data. Second, we coordinate this data through a progressive curriculum jointly optimizing composition, quality thresholds, and domain coverage across 20 trillion tokens. Third, to enable reasoning capabilities through scalable RL, we apply our proposed framework SnapPO for efficient optimization. Across benchmarks in English and Korean, Solar Open achieves competitive performance, demonstrating the effectiveness of this methodology for underserved language AI development.