Padding Tone: A Mechanistic Analysis of Padding Tokens in T2I Models Paper • 2501.06751 • Published 7 days ago • 31
LLMs Know More Than They Show: On the Intrinsic Representation of LLM Hallucinations Paper • 2410.02707 • Published Oct 3, 2024 • 48
Instructed to Bias: Instruction-Tuned Language Models Exhibit Emergent Cognitive Bias Paper • 2308.00225 • Published Aug 1, 2023