Upload README.md with huggingface_hub

README.md

@@ -1,30 +1,77 @@
-# A2 Pretrained Model
+---
+license: apache-2.0
+tags:
+  - robotics
+  - manipulation
+  - grasp
+  - lerobot
+  - clip
+---
 
-Pretrained ViLGP3D model for 6-DOF grasp pose selection in tabletop manipulation.
+# A2 Pretrained Policy
 
-## Model Architecture
+Pretrained ViLGP3D policy for 6-DOF grasp and place tasks in tabletop manipulation.
 
-- **Network**: CLIPAction (CLIP-based action selection with cross-attention)
-- **Width**: 768
-- **Layers**: 1
-- **Heads**: 8
-- **Action Dim**: 7 (xyz + quaternion)
-- **Features**: RoPE (Rotary Position Encoding)
+## Model Description
+
+This model uses CLIP-based cross-attention for selecting grasp and place poses from candidates generated by GraspNet/PlaceNet.
+
+## Files
+
+- `sl_checkpoint_199.pth`: Trained policy weights (ViLGP3D fusion network)
+- `checkpoint-rs.tar`: GraspNet checkpoint for grasp candidate generation
 
 ## Usage
 
+### With lerobot_policy_a2
+
 ```python
-from lerobot_policy_a2 import A2Policy, A2Config
+from lerobot_policy_a2 import A2Policy
 
 # Load pretrained model
 policy = A2Policy.from_pretrained("dgrachev/a2_pretrained")
+
+# Use for grasp prediction
+action, info = policy.predict_grasp(
+    color_images={"front": rgb_image},
+    depth_images={"front": depth_image},
+    point_cloud=point_cloud,
+    lang_goal="grasp a round object"
+)
+```
+
+### With LeRobot A2 Environment
+
+```bash
+# Data collection
+A2_DISABLE_EGL=true uv run python -m lerobot.envs.a2_collect     --policy a2     --hf_repo dgrachev/a2_pretrained     --task grasp     --num_episodes 100
+
+# Benchmark evaluation
+A2_DISABLE_EGL=true uv run python -m lerobot.envs.a2_benchmark     --task grasp     --policy a2     --hf_repo dgrachev/a2_pretrained
 ```
 
-## Training Data
+## Training Details
 
-Trained on simulated tabletop grasping with UR5e robot and Robotiq gripper.
+- **Architecture**: ViLGP3D with CLIP ViT-B/32 backbone
+- **Hidden dim**: 768
+- **Attention heads**: 8
+- **Position encoding**: Rotary Position Encoding (RoPE)
+- **Training data**: Tabletop manipulation demonstrations
 
-## Related
+## Related Resources
 
-- Environment: Install with `pip install lerobot[a2]`
-- Assets: [dgrachev/a2_assets](https://huggingface.co/datasets/dgrachev/a2_assets)
+- [lerobot_policy_a2](https://github.com/dgrachev/lerobot_policy_a2) - Policy package
+- [lerobot_grach0v](https://github.com/grach0v/lerobot) - LeRobot fork with A2 environment
+- [a2_assets](https://huggingface.co/datasets/dgrachev/a2_assets) - Environment assets
+
+## Citation
+
+```bibtex
+@misc{a2_policy,
+  author = {Denis Grachev},
+  title = {A2 Policy: CLIP-based 6-DOF Grasp and Place Policy},
+  year = {2025},
+  publisher = {HuggingFace},
+  url = {https://huggingface.co/dgrachev/a2_pretrained}
+}
+```