4+ hour, 44+ min ago  (309+ words) Apple Machine Learning Research Learning Structured Reasoning via Tractable Trajectory Control Large language models can exhibit emergent reasoning behaviors, often manifested as recurring lexical patterns (e. g. , "wait," indicating verification). However, complex reasoning trajectories remain sparse in unconstrained sampling, and standard RL…...

2+ mon, 3+ day ago  (134+ words) Authors Haoqiang Kang, Yizhe Zhang, Nikki Lijing Kuang, Nicklas Majamaki, Navdeep Jaitly, Yi-An Ma, Lianhui Qin Thinking into the Future: Latent Lookahead Training for Transformers March 25, 2026research area Methods and Algorithms Workshop at ICLR This paper was accepted at the Workshop…...

2+ mon, 1+ week ago  (757+ words) To accelerate research in efficient sequence modeling and enable researchers and practitioners to explore new nonlinear RNN models at scale, the Para RNN codebase has been released as an open-source framework for automatic training-parallelization of nonlinear RNNs. The computational cost…...

2+ mon, 3+ week ago  (250+ words) Apple Machine Learning Research A Theoretical Framework for Acoustic Neighbor Embeddings This paper provides a theoretical framework for interpreting acoustic neighbor embeddings, which are representations of the phonetic content of variable-width audio or text in a fixed-dimensional embedding space. A…...

3+ mon, 2+ hour ago  (12+ words) Personalized Group Relative Policy Optimization for Heterogenous Preference Alignment'Apple Machine Learning Research...

3+ mon, 3+ day ago  (284+ words) machinelearning. apple. com Policy gradient algorithms have driven many recent advancements in language model reasoning. An appealing property is their ability to learn from exploration on their own trajectories, a process crucial for fostering diverse and creative solutions. As we…...