Organization collection

nanoswe

Datasets published by this organization.

Organization inventory

DatasetAccessTasks
nanoswe/swe-bench-verified

SWE-bench Verified as used for nanoswe evals: 17 of the 500 instances are dropped because they do not run on our internal cluster, leaving a fixed 483-instance subset. Tasks are the standard SWE-bench Verified instances (not re-hosted); full agent trajectories are attached to the leaderboard entries.

Public
1

1 dataset