Yang Yue
Also known as 乐洋
Yang Yue is a PhD student at Tsinghua University in Gao Huang's research group, and their papers study how language models reason. A 2025 paper that Yang Yue co-wrote finds that current reinforcement learning training does not elicit fundamentally new reasoning patterns, and it received a Best Paper Runner-up award at NeurIPS 2025. In January 2026, the group's paper The Flexibility Trap argued that arbitrary-order generation can limit reasoning potential in diffusion language models. That paper won the Outstanding Paper Award at ICML 2026. Earlier, they co-wrote EfficientTrain++, a curriculum-learning paper for visual backbones, which appeared in IEEE TPAMI in 2024.
- Works
- 4
Publication numbers from OpenAlex, as of 15 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.
Timeline
-
2024
EfficientTrain++ curriculum learning
They co-write EfficientTrain++, a paper on curriculum learning for efficient visual backbone training, which appears in IEEE TPAMI in 2024.
-
April 2025
Limits of RL for reasoning
A paper they co-write, submitted to arXiv in April 2025, finds that the current training setup for reinforcement learning with verifiable rewards does not elicit fundamentally new reasoning patterns.
-
2025
NeurIPS Best Paper Runner-up
Their reinforcement learning paper receives the Best Paper Runner-up award at NeurIPS 2025.
-
January 2026
The Flexibility Trap
The Flexibility Trap, which they co-author, is submitted to arXiv in January 2026 and argues that arbitrary-order generation can limit reasoning potential in diffusion language models.
-
2026
ICML Outstanding Paper Award
Their co-authored paper, The Flexibility Trap, receives the Outstanding Paper Award at ICML 2026.
-
Since 2022
PhD study in Gao Huang's group
They begin PhD study at Tsinghua University in 2022 and remain a PhD student in Gao Huang's group.
Sources and credits
Photo: We have no published photo of Yang Yue yet. Send us a free-licensed photo.