The University of Washington published a review of October research, with particular attention given to two areas: compact artificial intelligence language models and the organization of DNA in stem cells.
In the field of AI, the team presented "micro language models" with 8–30 million parameters. These models are small enough to run on devices like smartwatches and provide instant response.
The model begins responding immediately, and after four to eight words a connected cloud LLM completes the answer. Errors in 4–8% of cases are corrected seamlessly.
In a user study, nine out of ten participants preferred such a hybrid system to a conventional cloud response with latency.
The models were presented on 27 October at the EMNLP 2026 conference. Among the authors are Wen Cheng, Tuochao Chen, Luke Zettlemoyer and Shyam Gollakota.
In parallel, UW biologists studied how the brain stem cells of fruit fly larvae organize DNA. Centromeres are usually located in the upper part of the nucleus.
The researchers showed that proteins outside and inside the nucleus jointly control this position. The work was published on 14 October in the Journal of Cell Biology.
When the positioning of centromeres is disrupted, the location of other genes also changes. The scientists used fluorescence microscopy and machine learning for measurements.
How do these mechanisms affect brain development as a whole? Studying the organization of DNA in stem cells helps to better understand stem cell regulation and brain formation.
The lead author is postdoc Jennifer Taylor, and the co-author is Clemens Cabernard.
