yoonholee/terminalbench-trajectories
Viewer • Updated • 52.1k • 3.22k • 18
Ana Do
anaexplores
AI & ML interests
Agent evaluation and reliability. Repeatability of benchmark results, ICC, pass^k, what a 66 task benchmark can and cannot tell apart. Small reproducible studies on public data.
Recent Activity
liked a dataset 12 days ago
yoonholee/terminalbench-trajectoriesOrganizations
None yet
Duplicated trials in 11 configurations, 4,952 rows with empty trial_id
#2 opened 12 days ago
by anaexplores
liked a
dataset 12 days ago