Datasets:

agent-perf-bench
/

AgentPerfBench

Removes raw MSE runner JSON payloads from the Hugging Face dataset repository and keeps mse_validation as a clean tabular subset. Raw aggregate/per-turn JSON artifacts remain in R2 and are referenced by raw_json_r2_uri / per_turn_json_r2_uri in summary.parquet and manifest.csv.\n\nConstraint: Submission dataset should expose loadable filtered rows, not a large tree of raw runner JSONs.\nRejected: Keeping raw JSONs in HF | makes mse_validation unlike the other curated configs and clutters the dataset surface.\nConfidence: high\nScope-risk: narrow\nDirective: Keep raw MSE artifacts in R2; keep HF mse_validation to parquet/CSV/notes only.\nTested: Verified no JSON files in upload payload; parsed README YAML; verified five configs; read regenerated summary.parquet; verified all 51 R2 artifact URIs exist.\nNot-tested: Did not run the Hugging Face worker locally because datasets is not installed here.

Keep MSE validation tabular in dataset repob6456568

booth-algo changed pull request status to merged 2 days ago

booth-algo deleted the refs/pr/8 ref 2 days ago

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment