Upload README.md with huggingface_hub

README.md

@@ -13,7 +13,7 @@ configs:
 - config_name: summary
   data_files:
   - split: train
-    path: roast_summary.csv
+    path: roast_summary.json
 - config_name: curves
   data_files:
   - split: train
@@ -31,8 +31,10 @@ and full attribution.
 
 ## Files
 - **roast_summary.csv** — one row per roast with the full technical summary (see field list below).
-- **roast_curves.parquet** — the same time-series subset in Parquet (this is the file the dataset viewer serves for the `curves` config; list columns are native here).
-- **roast_curves.json** — the time-series subset as a JSON array for readability/loading without extra deps (bean temperature at 1 Hz or per logged point, plus generated rate-of-rise): records of `{"roast_id", "t_s", "bean_temp_c", "ror_c_per_min", "equipment_type", "roast_level", "has_crash_flick"}`; null where a value doesn't apply (real curves have no generated RoR).
+- **roast_summary.json** — one row per roast with the full technical summary (JSON array; see field list below).
+- **roast_curves.json** — the time-series subset as a JSON array (bean temperature at 1 Hz or per logged point, plus generated rate-of-rise): records of `{"roast_id", "t_s", "bean_temp_c", "ror_c_per_min", "equipment_type", "roast_level", "has_crash_flick"}`; null where a value doesn't apply (real curves have no generated RoR).
+
+Both viewer-listed files are JSON on purpose: the dataset viewer routed this mixed-format repo's configs through its CSV reader (breaking the curves config), and a single format keeps every config loadable via `load_dataset("luispoveda93/coffee-roast-profiles", "summary" | "curves")`.
 - **validation_report.md** — synthetic-distribution check against the Roastetta n=3,286 aggregate.
 - **build_dataset.py** — the deterministic generator (seed=42) that produced the synthetic rows.