A Self-Pruning Transformer: Extreme KV-Cache Compression with Universal Attention · Decoded

[data-reveal]{opacity:1!important;clip-path:none!important;transform:none!important}[data-bar],[data-barx]{transform:none!important}[data-draw]{stroke-dashoffset:0!important}.op-ledger-row::before{transform:none!important}.op-ledger-num{opacity:1!important;transform:none!important}

Not decoded yet

in the queue

This paper is waiting its turn. The original abstract is at the bottom of the page.

From the paper