Create README.md

README.md

@@ -0,0 +1,34 @@
+# Diffusion Policy for Pick-and-Place (SO-100)
+
+## Overview
+This model is a Diffusion Policy trained using Hugging Face LeRobot for a single-task pick-and-place manipulation problem.
+
+## Robot
+- SO-100 follower arm
+- Front camera
+- Wrist camera
+
+## Task
+Pick the blue cube from one of four predefined locations and place it into a fixed target box.
+
+## Dataset
+- 40 teleoperated demonstrations
+- Dual-camera observations
+- Dataset:
+  Eshwar-2123/pick_place_blue_cube_box_clean_v1_20260706_125359
+
+## Training
+- Policy: Diffusion Policy
+- Steps: 30,000
+- Batch size: 16
+- GPU: NVIDIA RTX 5080
+
+## Results
+- Lowest validation loss: 0.037 (10k steps)
+- Final training loss: 0.0036
+- Final validation loss: 0.1303
+
+## Usage
+
+```python
+from lerobot.policies import DiffusionPolicy