Update README.md

README.md

@@ -36,19 +36,12 @@ augmented_grids[0] = [[0, 1, 0], [1, 0, 1], [0, 1, 0]]  # Example 3x3 grid
 ```
 
 ### Usage
-Load and use for AE pretraining:
+Load and use for pretraining:
 ```python
 from datasets import load_dataset
 ds = load_dataset("JDWebProgrammer/arc-agi-augmented")
-grids = ds['train']['augmented_grids']  # List of all grids
+grids = ds['augmented_grids']  # List of all grids
 
-# Example: Batch grids for AE
-def grid_to_tensor(grid):
-    h, w = len(grid), len(grid[0])
-    return torch.tensor(grid, dtype=torch.float).view(1, -1) / 10.0  # Normalize 0-1
-
-batch = torch.cat([grid_to_tensor(g) for g in grids[:32]])  # Batch of 32
-# Feed to AE: z = ae.encode(batch); recon = ae.decode(z)
 ```
 
 Ideal for: