Update README.md

README.md

@@ -6,33 +6,30 @@ emoji: 🏢
 colorFrom: purple
 colorTo: blue
 sdk_version: "1.33.0"
-python_version: "3.10
-
+python_version: "3.10"
+app_file: app.py
+---
 MindsEye Dataset Admin Console
 
-What this Space does
+This Space:
 
-Lists dataset configs and splits using HF Dataset Server
+Queries HF Dataset Server for splits/configs and first rows
 
-Previews first rows for a chosen config/split
+Uploads a CSV delta and commits it into your dataset repo under data/uploads/
 
-Uploads a CSV delta file and commits it into the dataset repo under data/uploads/
+Required secret
 
-Validates basic CSV header structure before uploading
+Go to your Space:
 
-Required Secret
-Set a Space Secret:
+Settings → Secrets → New secret
 
 Name: HF_TOKEN
 
-Value: (your Hugging Face access token)
+Value: your Hugging Face access token
 
-Where the token goes (exactly)
-Hugging Face Space → Settings → Secrets → New secret
-Name: HF_TOKEN
-Value: <PASTE TOKEN HERE>
+The app reads it via os.getenv("HF_TOKEN").
+
+Notes
+
+If your dataset is gated, your token must have access. If the dataset fails Parquet conversion, you likely need multiple configurations (e.g. nodes and runs) so HF doesn’t try to cast them into one schema.
 
-Notes on your dataset Parquet error
-Your dataset repo likely contains multiple CSVs under one config/split where columns differ (nodes vs runs).
-Fix is: define multiple configurations in the dataset repo (nodes, runs) OR separate into different splits.
-This Space helps you upload deltas safely, but the dataset repo still needs correct configuration.