fix stride unique samplin

src/lerobot/policies/rlearn/modeling_rlearn.py

@@ -983,33 +983,44 @@ class RLearNPolicy(PreTrainedPolicy):
             ep_length = ep_end - ep_start
             episode_lengths.append(ep_length)
             
-            # Choose random anchor - need enough frames before for stride sampling
+            # Proper window-relative stride sampling within available frames
-            # For T=16 and stride=2, we need frames [anchor-30, anchor-28, ..., anchor-2, anchor] 
             stride = self.config.temporal_sampling_stride
-            min_anchor = (T - 1) * stride  # Need (T-1)*stride frames before anchor
+            # Ensure we have room for T frames at given stride; shrink stride if needed
-            max_anchor = max(min_anchor, ep_length - 1)
+            if available_T <= 1:
-            anchor = torch.randint(min_anchor, max_anchor + 1, (1,)).item()
+                effective_stride = 1
-            anchor_positions.append(anchor)
+            else:
-            
+                effective_stride = max(1, min(stride, (available_T - 1) // max(T - 1, 1) if (T - 1) > 0 else 1))
-            # Build window indices with configurable stride sampling and reflection padding
+            min_anchor_in_window = (T - 1) * effective_stride
+            max_anchor_in_window = max(min_anchor_in_window, available_T - 1)
+            anchor_in_window = torch.randint(min_anchor_in_window, max_anchor_in_window + 1, (1,)).item()
+
+            # Convert window-anchor to episode-anchor (absolute frame index within episode)
+            cur_frame_idx = frame_indices[b_idx].item()
+            anchor_abs = cur_frame_idx + (anchor_in_window - (available_T - 1))
+            anchor_abs = int(max(0, min(anchor_abs, ep_length - 1)))
+            anchor_positions.append(anchor_abs)
+
+            # Build window indices with stride and reflection within [0, available_T)
             window_indices = []
-            frame_indices_for_progress = []  # Track actual frame positions for progress
+            frame_indices_for_progress = []  # Episode-relative absolute indices for progress
             had_oob = False
-            # Sample with stride: [anchor-(T-1)*stride, anchor-(T-2)*stride, ..., anchor-stride, anchor]
             for i in range(T):
-                delta = -(T - 1 - i) * stride  # Work backwards from anchor with stride spacing
+                delta = -(T - 1 - i) * effective_stride
-                idx = anchor + delta
+                w_idx = anchor_in_window + delta
-                actual_frame_idx = idx  # Store the actual frame index before reflection
+                if w_idx < 0:
-                if idx < 0:
+                    w_idx = -w_idx
-                    idx = -idx  # Reflect at start
                     had_oob = True
-                elif idx >= ep_length:
+                elif w_idx >= available_T:
-                    idx = 2 * (ep_length - 1) - idx  # Reflect at end
+                    w_idx = 2 * (available_T - 1) - w_idx
                     had_oob = True
-                window_indices.append(min(idx, available_T - 1))
+                w_idx = max(0, min(w_idx, available_T - 1))
-                # For reflected indices, use the reflected position for progress
+                window_indices.append(w_idx)
-                frame_indices_for_progress.append(idx)
+
-            
+                # Map window index back to episode-relative absolute frame index
+                abs_idx = cur_frame_idx + (w_idx - (available_T - 1))
+                abs_idx = int(max(0, min(abs_idx, ep_length - 1)))
+                frame_indices_for_progress.append(abs_idx)
+
             if had_oob:
                 oob_count += 1
                 
@@ -1021,7 +1032,7 @@ class RLearNPolicy(PreTrainedPolicy):
             # DEBUG: Check if stride sampling is producing different frames
             if torch.rand(1).item() < 0.1 and b_idx == 0:  # Debug first sample occasionally
                 print(f"\n🔍 STRIDE SAMPLING DEBUG (Sample {b_idx}):")
-                print(f"Episode length: {ep_length}, Anchor: {anchor}")
+                print(f"Episode length: {ep_length}, Anchor(abs): {anchor_abs}, Anchor(win): {anchor_in_window}, eff_stride: {effective_stride}")
                 print(f"Window indices: {window_indices[:5]}...{window_indices[-5:]}")  # First and last 5
                 print(f"Frame indices for progress: {frame_indices_for_progress[:5]}...{frame_indices_for_progress[-5:]}")