Loading activity…
@tatsu_hashimoto
Assistant Prof at Stanford CS, member of @stanfordnlp and statsml groups; Formerly at Microsoft / postdoc at Stanford CS / Stats.
AI
Hinted scaling laws use easier benchmark versions; researchers report median forecasting error fell from 21 to 4.3 percentage points.
AI
A 35-million-parameter transformer reportedly kept lowering its loss after sharpness crossed a predicted stability threshold in a CS 312 experiment.
AI
A developer credits retrieval-augmented n-gram hints for the claimed 83.2% cut in time.
XAI
A co-instructor says seven exams account for 85% of the grade, with in-person discussion assessments making up 15%.
