Spurious Rewards Paradox: Mechanistically Understanding How RLVR Activates Memorization Shortcuts in LLMs
Paper • 2601.11061 • Published • 7
Ruizhe Li
rzdiversity
·
AI & ML interests
Mechanistic Interpretability, Multimodal LLMs
Organizations
None yet
authored a
paper 3 months ago
submitted a
paper to Daily Papers 3 months ago
authored a
paper 11 months ago
authored 7
papers over 1 year ago
Latent Space Factorisation and Manipulation via Matrix Subspace Projection
Paper • 1907.12385 • Published • 1
Large Language Models are Efficient Learners of Noise-Robust Speech Recognition
Paper • 2401.10446 • Published
It's Never Too Late: Fusing Acoustic Information into Large Language Models for Automatic Speech Recognition
Paper • 2402.05457 • Published
GenTranslate: Large Language Models are Generative Multilingual Speech and Machine Translators
Paper • 2402.06894 • Published • 1
Adaptive Retrieval-Augmented Generation for Conversational Systems
Paper • 2407.21712 • Published • 2
Detecting Machine-Generated Texts: Not Just "AI vs Humans" and Explainability is Complicated
Paper • 2406.18259 • Published
Overcoming Catastrophic Forgetting by Exemplar Selection in Task-oriented Dialogue System
Paper • 2405.10992 • Published
authored a
paper almost 2 years ago