Hill-Climbing Skills: Improve Agents Without Changing the Model — Shubhankar Srivastava, Browserbase
YouTube
Hill-Climbing Skills: Improve Agents Without Changing the Model — Shubhankar Srivastava, Browserbase
An agent succeeds at a flight search, then fails on its next iteration. Shubhankar Srivastava uses that uneven progress to show how browser agents can learn a task without changing model weights. His AutoBrowse loop pairs a doer that operates the browser…