Ruslan Salakhutdinov

RLAD: Training LLMs to Discover Abstractions for Solving Reasoning Problems

InSTA: Towards Internet-Scale Training For Agents

Accelerating Diffusion Planners in Offline RL via Reward-Aware Consistency Trajectory Distillation

Joint Distillation for Fast Likelihood Evaluation and Sampling in Flow-based Models

Large Language Models as Improvement Operators: Better Reasoning by Iteration

Learning to Reason on Hard Problems with Privileged On-Policy Exploration

Adaptively Labeling Vision Datasets Via Instance-Level Retrieval

Verlog: An Efficient Synchronized Multi-turn RL Framework for LLM Agents

Can Large Reasoning Models Self-Train?

Optimizing Test-Time Compute via Meta Reinforcement Finetuning

Retrospective In-Context Learning for Temporal Credit Assignment with Large Language Models