21 days ago
-
v18: four-task benchmark (shared seed-42 split, HGB baselines), splits/results/benchmark.md published
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload benchmark.md with huggingface_hub
-
Upload benchmark_results.json with huggingface_hub
-
Upload benchmark_splits.json with huggingface_hub
-
v17: RoR backfill for all curves, equipment/level backfill from summary, README null-policy section
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload roast_curves.json with huggingface_hub
-
v16: enrichment — derived features (features config), per-origin reference table (reference config), roast_id fix on real curves
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload roast_reference.json with huggingface_hub
-
Upload roast_features.json with huggingface_hub
-
Upload roast_curves.json with huggingface_hub
25 days ago
-
v11: curves config path -> roast_curves.json
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Remove parquet: the viewer resolved one CSV builder per mixed-format dataset; viewer files are now JSON only
-
v10: all viewer-listed files JSON (summary+curves) for single-builder viewer resolution
-
Remove csv: the viewer resolved one CSV builder per mixed-format dataset; summary now ships as roast_summary.json (identical content)
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload roast_summary.json with huggingface_hub
-
v9: curves config on parquet twin (viewer-native); json kept for humans
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload roast_curves.parquet with huggingface_hub
-
v8: float t_s in curves (viewer type-mismatch fix) + load_dataset self-check before push
-
Upload validation_report.md with huggingface_hub
-
Upload roast_curves.json with huggingface_hub
-
v7: curves as .json array for viewer compatibility; README config paths updated
-
Remove .jsonl: datasets 4.x parses .jsonl via the CSV builder, which broke the dataset viewer; curves now ship as roast_curves.json
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload roast_curves.json with huggingface_hub
-
v6: dataset viewer configs in README YAML, uniform-schema curve records
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload roast_curves.jsonl with huggingface_hub
-
Add deterministic generator script (seed=42) used to build the synthetic portion
-
Upload README.md with huggingface_hub
-
Upload validation_report.md with huggingface_hub
-
Upload roast_curves.jsonl with huggingface_hub
-
Upload roast_summary.csv with huggingface_hub
-
initial commit
