change explanations

lerobot/common/policies/vqbet/modeling_vqbet.py

@@ -449,9 +449,9 @@ class VQBeTHead(nn.Module):
             torch.arange(self.config.vqvae_groups, device=device).unsqueeze(0),
             sampled_centers,
         )
-        # Use advanced indexing to sample the values
+        # Use advanced indexing to sample the values (Extract the only offsets corresponding to the sampled codes.)
         sampled_offsets = cbet_offsets[indices]
-        # Extract the only offsets corresponding to the sampled codes.
+        # Then, sum the offsets over the RVQ layers to get a net offset for the bin prediction
         sampled_offsets = sampled_offsets.sum(dim=1)
         # Get the centroids of each layer to pass it through RVQ decoder
         centers = self.vqvae_model.draw_code_forward(sampled_centers).view(