AAlex ShawFix formatting across entire repo.
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
Create a terminal-bench CLI, improve module packaging, create lockfiles, and integrate dataset registry. (#298) * Create a robust CLI. * Create a basic CLI and tb.lock file for reproducibility. * Record if the install is editable (bc it presents reproducibility risks). * Update dataset location and enable dataset registering on supabase. * Fix the dataset paths. * Rename param. * Change experiments to runs as default output dir. * Update the run harness script. * Make supabase primary dep. * Split out the shared folder into a dataset folder. * Fix task references. * Save invocation. Refactor tb CLI. * Create runs commands. * Alexgshaw/tbext 107 GitHub as a registry (#310) * Fix setting format. * Githuh client. * Update the registry client. * Modify CLI. * Fix the client. * Implement the CLI. * Fix datasets. * Remove. * Clone the branch directly. * Fix the client. * Remove url and checksum for now. * Add defaults for backwards compatibility. * Save registry metadata for reproducibility. * Pass registry to the harness. * main registry. * Add TODO. * Restructure agents and name. * Cahnge task-ids to task-id * get_name -> name * Fix imports. | 1 年前 | |
Fix formatting across entire repo. | 1 年前 | |
WYSIWYG tasks (#343) * Copy over all dependencies to the task files themselves. * Get rid of defaults, enviornments, and scripts. * Get rid of test scripts and env name params from task files. * Refactor trial handler. * Fix build dir. * Remove build context dir from code. * Straggler. * Update the wizard to include the new files. * Remove the old wizard command. * Deprecate run_harness.py * Fix chess solution. * Fix run tests. * Fix simple sheets tests. * Restructure folders. * Add the fork task wizard * Make task id an arg. * Fix lines. * Add canary strings. * Remove old run_harness uses. * Fix modified tasks. * Fix run harness. * Modify to only expect tasks. * Update the registry file. * Increment version and reformat. | 1 年前 |