Mykola Pechenizkiy

Leave it to the Specialist: Repair Sparse LLMs with Sparse Fine-Tuning via Sparsity Evolution

Mutual Information Dynamics Learning: A New Paradigm for Unsupervised Reinforcement Learning

MuBench: Assessment of Multilingual Capabilities of Large Language Models Across 61 Languages

ProSocialAlign: Preference-Conditioned Test-Time Alignment in Language Models

ParaRater: Enhancing Cross-Lingual Transfer in Large Langauge Models with Meta-Learning

Think you have Solved Commonsense Reasoning? Try HellaswagUltra

MEAL: A Benchmark for Continual Multi-Agent Reinforcement Learning

Variational Inference based Probabilistic Prompt for Rehearsal-Free Continual Learning

Propensity Guided Transformer For Causal Effect Inference

Distill Not Only Data but Also Rewards: Can Smaller Language Models Surpass Larger Ones?

Training with Dynamic Sparse Heads as the Key to Effective Ensembling

Preference Controllable Reinforcement Learning with Advanced Multi-Objective Optimization

RuAG: Learned-rule-augmented Generation for Large Language Models

GasketRAG: Systematic Alignment of Large Language Models with Retrievers

HASARD: A Benchmark for Vision-Based Safe Reinforcement Learning in Embodied Agents

Customizing Reinforcement Learning Agent with Multi-Objective Preference Control

Dynamic Sparse Training versus Dense Training: The Unexpected Winner in Image Corruption Robustness

GPTBIAS: A Comprehensive Framework for Evaluating Bias in Large Language Models

Assessing Vulnerabilities of Large Language Models to Social Bias Attacks