cs.IR · 2026-08-23 · No. 93
Information Retrieval, 2026-08-23.
1 new papers in cs.IR. Titles, authors,
abstracts. Links to arXiv. Want this in your inbox every morning? Subscribe →
01 — The papers
1 entries-
01
Daedalus-150M: A Convolution-Attention Hybrid Designed for CPU Inference
Christos Koutsiaris
cs.IR · cs.AI · cs.CL · cs.LG
Small language models are usually built like large ones and then squeezed onto a CPU afterwards. We did the opposite: we fixed the target first, one user, one token at a time, 4-bit weights, ordinary CPU, and chose the architecture to suit it. The result keeps full attention in only 6 of its 18 blocks. The other 12 use short convolutions whose memory is two timesteps wide no matter how long the conversation gets, so two thirds of the network...
This edition is part of The Daily Abstract — cs.IR archive. Subscribe to receive these in your inbox each morning, automatically translated to Spanish, with reply-to-PDF: arxivdaily.ignorelist.com.
#D99C5E. Built and served on an always-free VM. The masthead is set 14% letterspaced because newspapers do that and it works.