Mitigating Attention Sinks and Massive Activations in Audio-Visual Speech Recognition with LLMS

Add code
Oct 26, 2025

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: