Anirudh Goyal

On the Impossibility of Retrain Equivalence in Machine Unlearning

Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision

Large Language Models as Improvement Operators: Better Reasoning by Iteration

Metacognitive Reuse: Turning Recurring LLM Reasoning Into Concise Behaviors

HypoSpace: Evaluating LLM Creativity as Set-Valued Hypothesis Generators under Underdetermination

ASTRO: Teaching Language Models to Reason by Reflecting and Backtracking In-Context

Understanding and Enhancing Safety Mechanisms of LLMs via Safety-Specific Neuron

COrAL: Order-Agnostic Language Modeling for Efficient Iterative Refinement

Low Compute Unlearning via Sparse Representations

Masked Generative Priors Improve World Models Sequence Modelling Capabilities

On the Transfer of Object-Centric Representation Learning

Generalizing from SIMPLE to HARD Visual Reasoning: Can We Mitigate Modality Imbalance in VLMs?

Unnatural Languages Are Not Bugs but Features for LLMs

Instruct-SkillMix: A Powerful Pipeline for LLM Instruction Tuning

Narrowing the Focus: Learned Optimizers for Pretrained Models

AI-Assisted Generation of Difficult Math Questions

SWE-Search: Enhancing Software Agents with Monte Carlo Tree Search and Iterative Refinement

Bayesian-LoRA: LoRA based Parameter Efficient Fine-Tuning using Optimal Quantization levels and Rank Values trough Differentiable Bayesian Gates