AI for Scientific Search(arxiv.org)
108 points by omarsar | 26 comments
108 points by omarsar | 26 comments
121 points by milliondreams | 21 comments
77 points by alexmolas | 44 comments
525 points by weirdcat | 166 comments
arXiv paper introducing Diff Transformer, which computes attention as the difference of two softmax maps to cancel noise and improve long-context modeling and reduce hallucination.