Position-Aware Depth Decay Decoding ($D^3$): Boosting Large Language Model Inference Efficiency

Add code
Mar 11, 2025

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: