10 papers
One Block Ahead: Promoting Sampler State in Diffusion Language Models
Yangyi Shen, Pin-Jui Ku, Nicolai Oswald +6
Block-diffusion language models denoise a whole block of token positions in parallel so each forward pass can emit many tokens. However, blocks are still denoised in strict sequenc…
Bandits with Multiple Optimal Arms: Minimax Regret and Non-Adaptivity
Kaixuan Ji, Qiwei Di, Qingyue Zhao +2
We study multi-armed bandits (MAB) with multiple optimal arms, motivated by the fact that many practical decision making problems admit multiple correct answers. For -armed band…
On the Optimal Sample Complexity of Offline Multi-Armed Bandits with KL Regularization
Kaixuan Ji, Qiwei Di, Heyang Zhao +2
Kullback-Leibler (KL) regularization is widely used in offline decision-making and offers several benefits, motivating recent work on the sample complexity of offline learning with…
Transformers Trained via Gradient Descent Can Provably Learn a Class of Teacher Models
Chenyang Zhang, Qingyue Zhao, Quanquan Gu +1
Transformers have achieved great success across a wide range of applications, yet the theoretical foundations underlying their success remain largely unexplored. To demystify the s…
Near-Optimal Regret for KL-Regularized Multi-Armed Bandits
Kaixuan Ji, Qingyue Zhao, Heyang Zhao +2
Recent studies have shown that reinforcement learning with KL-regularized objectives can enjoy faster rates of convergence or logarithmic regret, in contrast to the classical $\sqr…
Robust Layerwise Scaling Rules by Proper Weight Decay Tuning
Zhiyuan Fan, Yifeng Liu, Qingyue Zhao +2
Empirical scaling laws prescribe how to allocate parameters, data, and compute, while maximal-update parameterization (P) enables learning-rate transfer across widths by equaliz…
Policy Frameworks for Transparent Chain-of-Thought Reasoning in Large Language Models
Yihang Chen, Haikang Deng, Kaiqiao Han +1
Chain-of-Thought (CoT) reasoning enhances large language models (LLMs) by decomposing complex problems into step-by-step solutions, improving performance on reasoning tasks. Howeve…
Global Convergence and Rich Feature Learning in -Layer Infinite-Width Neural Networks under P Parametrization
Zixiang Chen, Greg Yang, Qingyue Zhao +1
Despite deep neural networks' powerful representation learning capabilities, theoretical understanding of how networks can simultaneously achieve meaningful feature learning and gl…
Towards a Sharp Analysis of Offline Policy Learning for -Divergence-Regularized Contextual Bandits
Qingyue Zhao, Kaixuan Ji, Heyang Zhao +2
Many offline reinforcement learning algorithms are underpinned by -divergence regularization, but their sample complexity *defined with respect to regularized objectives* still…
Towards the Fundamental Limits of Knowledge Transfer over Finite Domains
Qingyue Zhao, Banghua Zhu
We characterize the statistical efficiency of knowledge transfer through samples from a teacher to a probabilistic student classifier with input space over labels…