Efficiently combining experts improves large ai model performance and speed
Turing LLM Post-Training: Calibrated Distillation and Efficient Dynamic-MoE Execution
Machine LearningArtificial Intelligence
Summary
Large AI models often include many specialized parts called experts that perform different tasks. This paper shows a way to combine these experts after training so the model works well across multiple tasks without slowing down too much. The authors created a method to carefully balance the influence of each expert and speed up how the combined model runs. This approach leads to better results on difficult math and coding tests while also running much faster than before.
What this means in practice
- •For ai development teams: Integrate independently trained specialist models into one efficient large language model without losing specialist performance.
- •For cloud infrastructure engineers: Reduce the time and computing resources needed to consolidate mixture-of-experts models during deployment by using faster execution strategies.
Authors
Yue He, Jiahui Hu, Shaman Tang, Hua Zhou, Qiman Wu, Menglin Li, Yizhao Wang, Yuheng Zhang, Da Zhu, Zhou Wang, Anhua Liu, Hanlin Chen, Mingyuan Wang, Rui Xin, Yuhua Wei, Shuang Su, Siqi Liu, Kai Yang, Weichao Huang, Hang Zhang, Honggou Yang, Xianming Liu
Abstract
We study post-training for dynamically routed mixture-of-experts (MoE) models, focusing on integrating independently optimized specialists while keeping consolidation efficient. Our approach combines calibrated multi-teacher on-policy distillation (C-MOPD), which matches each selected teacher on a local Top-$K$ support with teacher-wise loss normalization and clipping, with an execution path that uses post-capacity bounded-width dispatch, final-route replay, and asynchronous rollout--training overlap. On Turing-20B-A2B, the selected model achieves a four-domain normalized integration score of 0.9988 and retains at least 90.16\% of the measured specialist gain in every domain. At matched $K=32$, the full calibration objective scores 0.9962, compared with 0.9724 without calibration and 0.9834 without clipping. At a matched 8K response cap, the model exceeds both shared SFT and Qwen3-8B on all four reasoning-enabled mathematics and code benchmarks (unweighted mean: 85.92, versus 80.33 for Qwen3-8B and 74.21 for SFT), and improves over SFT on ten of twelve broader evaluation metrics. In a PPU-based profile, the combined execution configuration reduces consolidation step time from 398.7 to 139.3 seconds, an observed 2.86$\times$ speedup over serial execution. These results demonstrate strong specialist integration alongside more efficient consolidation, while reflecting the complete training and execution framework rather than any single component.