Hunyuan-Large: An Open-Source MoE Model with 52 Billion Activated Parameters by Tencent

TL;DR

Hunyuan-Large is the largest open-source Transformer-based mixture of experts model with 389 billion parameters, demonstrating superior performance across multiple benchmarks and introducing novel training techniques and insights into MoE scaling laws.

Contribution

It introduces Hunyuan-Large, the largest open-source MoE model, with innovative training strategies and comprehensive analysis of MoE scaling laws and learning rate schedules.

Findings

Outperforms LLama3.1-70B across various tasks

Comparable to LLama3.1-405B in performance

Provides insights into MoE scaling and optimization

Abstract

In this paper, we introduce Hunyuan-Large, which is currently the largest open-source Transformer-based mixture of experts model, with a total of 389 billion parameters and 52 billion activation parameters, capable of handling up to 256K tokens. We conduct a thorough evaluation of Hunyuan-Large's superior performance across various benchmarks including language understanding and generation, logical reasoning, mathematical problem-solving, coding, long-context, and aggregated tasks, where it outperforms LLama3.1-70B and exhibits comparable performance when compared to the significantly larger LLama3.1-405B model. Key practice of Hunyuan-Large include large-scale synthetic data that is orders larger than in previous literature, a mixed expert routing strategy, a key-value cache compression technique, and an expert-specific learning rate strategy. Additionally, we also investigate the scaling laws and learning rate schedule of mixture of experts models, providing valuable insights and guidances for future model development and optimization. The code and checkpoints of Hunyuan-Large are released to facilitate future innovations and applications. Codes: https://github.com/Tencent/Hunyuan-Large Models: https://huggingface.co/tencent/Tencent-Hunyuan-Large