Create README.md

README.md

@@ -1,35 +1,21 @@
----
-dataset_info:
-  features:
-  - name: timestamp
-    dtype: float64
-  - name: frame_index
-    dtype: int64
-  - name: episode_index
-    dtype: string
-  - name: task_index
-    dtype: string
-  - name: index
-    dtype: int64
-  - name: next.done
-    dtype: bool
-  - name: next.reward
-    dtype: float64
-  - name: observation.state
-    list: float64
-  - name: observation.image
-    dtype: string
-  - name: action
-    list: float64
-  splits:
-  - name: train
-    num_bytes: 961
-    num_examples: 3
-  download_size: 5575
-  dataset_size: 961
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
----
+## 📊 Interactive Data Viewer
+
+Explore real robot demonstration samples from the **Piper Arm** dataset directly in your browser! This interactive viewer lets you browse episodes, inspect images, actions, and robot states.
+
+<iframe
+  src="https://huggingface.co/datasets/ISdept/piper_arm/embed"
+  width="100%"
+  height="600px"
+  frameborder="0"
+  style="border: 1px solid #ddd; border-radius: 8px;"
+></iframe>
+
+> 🔍 What you can see:
+> - ✅ RGB images from front-facing camera (640×480)
+> - ✅ Action commands (6D: dx, dy, dz, droll, dpitch, dyaw)
+> - ✅ Robot joint states and end-effector pose
+> - ✅ Episode success signal (`done`, `reward`)
+> - 🔁 Navigate between samples using "Next" / "Prev" buttons
+> - 🧩 Filter by split: `train`, `test`, or `val`
+
+💡 Tip: Click any row to expand and view full details. Use the "Show more" button to see hidden fields.