5 papers
Entropy Regularization: A Free Correction to Cross-Entropy for Verified Demonstrations
Mihir Dhanakshirur, Adam Ousherovitch, Ambuj Tewari
Large language models are often post-trained on expert demonstrations using cross-entropy (CE), even when the downstream objective is not to imitate the demonstrated solution but t…
Direct Optimization of Generators for Search in Automated Theorem Proving
Adam Ousherovitch, Ambuj Tewari
Fine-tuned Large Language Models (LLMs) significantly advance Automated Theorem Proving (ATP), but are often deployed as guiding policies within tree search rather than for single-…
Terminal Shrinkage Averaging Reveals a Schedule-Estimator Interaction in LLM Pretraining
Adam Ousherovitch, Yixin Wang
Large language model (LLM) pretraining conventionally returns the raw final iterate. This couples two design choices: the learning-rate schedule that generates the parameter trajec…
Learning to Extrapolate to New Tasks: A Relational Approach to Task Extrapolation
Adam Ousherovitch, Yixin Wang
Modern learning systems excel at interpolation but struggle to generalize to unseen tasks outside the training distribution's support. This failure occurs even in simple settings,…
Compute Aligned Training: Optimizing for Test Time Inference
Adam Ousherovitch, Ambuj Tewari
Scaling test-time compute has emerged as a powerful mechanism for enhancing Large Language Model (LLM) performance. However, standard post-training paradigms, Supervised Fine-Tunin…