LooM: Loop-Invariant Memory for Causal Encoder-Decoder Language Models

In 30 Sekunden

Was passiert ist

LooM enables looped Transformer decoders to reuse cache across iterations, supporting variable compute budgets from 3 to 21 layer executions per token without memory growth.

Warum es zählt

Relevant for engineers optimizing inference cost and latency in encoder-decoder models where deployment-time compute budgets vary across different hardware or use cases.

Achtung

Paper is recent preprint with no reported implementation or benchmark comparisons against other looped Transformer approaches, limiting practical deployment guidance.