10 papers
Vocabulary Dropout for Curriculum Diversity in LLM Co-Evolution
Jacob Dineen, Aswin RRV, Zhikun Xu +1
Co-evolutionary self-play, where one language model generates problems and another solves them, promises curriculum learning without human supervision. The promise breaks down earl…
Mid-Training with Self-Generated Data Improves Reinforcement Learning in Language Models
Aswin RRV, Jacob Dineen, Divij Handa +4
The effectiveness of Reinforcement Learning (RL) in Large Language Models (LLMs) depends on the nature and diversity of the data used before and during RL. In particular, reasoning…
GuidedSampling: Steering LLMs Towards Diverse Candidate Solutions at Inference-Time
Divij Handa, Mihir Parmar, Aswin RRV +3
Repeated Sampling (RS) is a simple inference-time algorithm that has been shown to improve model performance on complex tasks. Although it is an effective way of scaling inference…
QA-LIGN: Aligning LLMs through Constitutionally Decomposed QA
Jacob Dineen, Aswin RRV, Qin Liu +8
Alignment of large language models (LLMs) with principles like helpfulness, honesty, and harmlessness typically relies on scalar rewards that obscure which objectives drive the tra…
PHANTOM RECALL: When Familiar Puzzles Fool Smart Models
Souradeep Mukhopadhyay, Rishabh Baral, Nimeesh Mahajan +5
Large language models (LLMs) such as GPT, Gemini, and Claude often appear adept at solving classic logic puzzles--but how much genuine reasoning underlies their answers? Recent evi…
ThinkTuning: Instilling Cognitive Reflections without Distillation
Aswin RRV, Jacob Dineen, Divij Handa +4
Recent advances in test-time scaling have led to the emergence of thinking LLMs that exhibit self-reflective behaviors and multi-step reasoning. While RL drives this self-improveme…