ICML
ICML, leading international academic conference in machine learning
Episodes
-
#5625: Omni and the Marble That Rolls UphillGoogle's Omni claims an intuitive grasp of gravity and momentum. Can a video model really learn physics, or just remember what physics looks like? -
#5805: Freeze the Model, Move the LabelsYour classifier isn't stale — your label set moved. Why retraining is cheap, and why not retraining might be smarter. -
#5466: Hebrew Words Hidden in English TextDaniel wants a classifier that spots Hebrew written in Latin letters — and it turns out nobody's built one. -
#5411: Fine-Tuning at 4-Bit vs 16-Bit: What It Really CostsQLoRA cuts fine-tuning VRAM 15x and cost up to 85% — but you pay in training time, quality, and safety alignment.