Waterloo's PAW compiles task specs into 23MB LoRA adapters a 600M-parameter model runs entirely offline.

Local AI inference at 32B-parameter quality, no cloud API required: University of Waterloo researchers released PAW on July 2, 2026, a system that compiles any natural-language task spec into a 23MB ...
Tech Times

Compile Once, Run Offline: New AI Method Matches 32B Models With a 23MB File

Local AI inference at 32B-parameter quality, no cloud API required: University of Waterloo researchers released PAW on July 2 ...
Werd I/O on MSNOpinion

Notable links: July 3, 2026

AI, surveillance, open tech, and news as a business.
New Haven Independent

From Sheff v. O’Neill To The 97th District Race

Wildaliz Bermúdez connects her state rep campaign to the landmark school-desegregation court case for which she served as one of 18 child plaintiffs.