[CL] EvoLM: In Search of Lost Language Model Training Dynamics

[Harvard & Stanford & EPFL]

https://arxiv.org/abs/2506.16029

真正的自由，是从不再解释开始。当你不再需要向世界证明什么时，你才真正开始拥有了自己的人生。

[CV] Align Your Flow: Scaling Continuous-Time Flow Map Distillation  

[NVIDIA]  

https://arxiv.org/abs/2506.14603

[LG] Taming Polysemanticity in LLMs: Provable Feature Recovery via Sparse Autoencoders  

[Yale University & Shanghai Jiao Tong University]  

https://arxiv.org/abs/2506.14002

[LG] Flat Channels to Infinity in Neural Loss Landscapes  

[EPFL & Flatiron Institute]  

https://arxiv.org/abs/2506.14951

[LG] GrokAlign: Geometric Characterisation and Acceleration of Grokking  

[Rice University & Brown University]  

https://arxiv.org/abs/2506.12284

真正的自由，从来不是拥有无限的选择权。

[CL] AceReason-Nemotron 1.1: Advancing Math and Code Reasoning through SFT and RL Synergy  

[NVIDIA]  

https://arxiv.org/abs/2506.13284

[LG] Random Matrix Theory for Deep Learning: Beyond Eigenvalues of Linear Models  

[Huazhong University of Science and Technolog & UC Berkeley]  

https://arxiv.org/abs/2506.13139

[LG] LiveCodeBench Pro: How Do Olympiad Medalists Judge LLMs in Competitive Programming?  

[New York University & Princeton University]  

https://arxiv.org/abs/2506.119