dataset readme

README.md

@@ -1,31 +1,114 @@
-# wam-0.1
+---
+license: other
+pretty_name: WAM-0.1
+tags:
+  - robotics
+  - panda
+  - robosuite
+  - robocasa
+  - h-jepa
+  - world-model
+---
 
-Panda action-conditioned dataset project for H-JEPA.
+# WAM-0.1 Dataset 
 
-The project has two learning levels:
+WAM-0.1 is an image-and-action dataset for learning Panda robot dynamics and world representations.
 
-- Level 0: input frame + one 7D Panda action -> observation frame
-- Level 1: input frame + action chunk -> future subgoal frame
+Created by **Atom Ctrl** and **Aditya Panigrahi**.
 
-The current approved test set contains 10 RoboSuite Panda tasks. Each task is tested in normal and randomized modes. Randomization changes the approved scene, camera, lighting, and object conditions while keeping the task controller and fixed white table rule intact.
+Website: https://atomctrl.com
 
-Level 1 is built from consecutive Level 0 transitions. It never mixes episodes or chooses a random endpoint:
+## Dataset overview
 
-```text
-Level 0: input[t] + action[t] -> observation[t]
-Level 0: observation[t] == input[t + 1]
-Level 1: input[t] + actions[t:t+8] -> observation[t+7]
-```
+The dataset contains two aligned learning levels.
 
-## Project commands
+![image](https://cdn-uploads.huggingface.co/production/uploads/68dc1570c7fc381a7bae2aaa/JxfpN9VA5M5vEnCuEKV-6.png)
 
-```bash
-source ~/robotics/roboenv/bin/activate
-cd ~/robotics/wam-0.1
-python scripts/show_plan.py
-python scripts/test_controller.py
-python scripts/test_two_levels.py
-python scripts/collect_batch.py --workers 2
-```
+### Level 0
 
-The two-level test writes only to `data/test_two_levels`. The production directory `data/final` remains empty until the batch collector is implemented and tested.
+One action predicts the next observation:
+
+`input image + 7D action -> next image`
+
+### Level 1
+
+An action chunk predicts a future subgoal:
+
+`start image + action chunk -> end image`
+
+Level 1 samples correspond directly to sequences of Level 0 transitions.
+
+## Data sources
+
+- Official RoboSuite Panda tasks
+- Randomized Panda trajectories
+- Official RoboCasa demonstrations
+- RGB and depth images
+- Robot state and Panda actions
+- Level 0 transitions and Level 1 action chunks
+
+## Visual examples
+
+### Normal trajectories
+
+
+![Normal trajectory](https://cdn-uploads.huggingface.co/production/uploads/68dc1570c7fc381a7bae2aaa/5NM8_MREYvVHvie2txJo-.png)
+
+### Randomized trajectories
+
+
+![Random trajectory](https://cdn-uploads.huggingface.co/production/uploads/68dc1570c7fc381a7bae2aaa/VvZIjkTKYBmxTBo_QuqHF.png)
+
+### RoboCasa trajectories
+
+
+![RoboCasa trajectory](https://cdn-uploads.huggingface.co/production/uploads/68dc1570c7fc381a7bae2aaa/R_Kg1TxijlVv9Ona0pqRt.png)
+
+### Additional examples
+
+
+![Additional trajectory](https://cdn-uploads.huggingface.co/production/uploads/68dc1570c7fc381a7bae2aaa/FECjQ5dwtG3gs9K2Rh4AW.png)
+
+## Core tasks
+
+- Lift
+- Stack
+- Door
+- Wipe
+- PickPlaceBread
+- PickPlaceMilk
+- PickPlaceCereal
+- PickPlaceCan
+- PickPlaceSingle
+- NutAssemblySquare
+
+## Dataset layout
+
+    data/
+    └── final/
+        ├── level_0/
+        │   ├── normal/
+        │   ├── random/
+        │   └── robocasa/
+        └── level_1/
+            ├── normal/
+            ├── random/
+            └── robocasa/
+
+## Intended use
+
+WAM-0.1 is intended for H-JEPA training, image-based world models, action-conditioned prediction, robot dynamics learning, subgoal prediction, and robotics planning research.
+
+## Download
+
+    hf download atomctrl/wam-0.1 --repo-type dataset --local-dir wam-0.1
+
+## Citation
+
+    @dataset{panigrahi_wam_0_1,
+      author    = {Aditya Panigrahi},
+      title     = {WAM-0.1: World Action Model Dataset},
+      year      = {2026},
+      publisher = {Atom Ctrl},
+      url       = {https://huggingface.co/datasets/atomctrl/wam-0.1}
+    }