Browserbase
@browserbase
Browser evals are only valuable if they hold up on the live internet. Most evals break with site updates, bot protection, and faulty graders.
We partnered with @hud_evals to write about what makes a good browser task that frontier labs actually want.
Read the full post:
We partnered with @hud_evals to write about what makes a good browser task that frontier labs actually want.
Read the full post:
hud@hud_evals · Sep 28Most browser evals look clean in a sandbox and fall apart on the live web. We wrote a walkthrough with @browserbase on how to write browser evals that hold up on the live web.
Sites update, graders reward faked outputs, and bot detection kill your runs before the agent even
Sites update, graders reward faked outputs, and bot detection kill your runs before the agent even
Open quoted post →
3 32