Aly M. Kassem
@_AKassem
Exploration over Exploitation. RA @Mila_Quebec. MSc @UWindsor. Interested in Adversarial attacks, security & reliability of LLMs
Finally, the crazy weeks of NeurIPS ddl, ICCV rebuttal and EMNLP ddl have passed. Now it's time to take a rest till Sep!
Very useful thread — unfortunately, I learned it the hard way.
complete blog post is now live! more tips, and reframed into 2-step process: (1) get accepted by page 1, (2) avoid rejection with the rest. download both PDFs too. linked in next.
Check out my mentee's latest work on LLM Router attack! First work on this topic to the best of our knowledge. Read our paper at: zhijing-jin.com/files/papers/2… Great job to @_AKassem!🎉 @MPI_IS @UofTCompSci
Router-LLMs promise to balance LLM cost & performance. But how robust are their routing decisions? 🤔 Our new paper analyzes their fragility, finding they often rely on category heuristics, especially for simple tasks & safety, by evaluating open/closed source routers 🧵