diff --git a/ml/src/trainers/tft.rs b/ml/src/trainers/tft.rs
index 2b55e9ed3..8844bb3f4 100644
--- a/ml/src/trainers/tft.rs
+++ b/ml/src/trainers/tft.rs
@@ -1461,7 +1461,7 @@ impl TFTTrainer {
             None
         };
 
-        for batch in val_loader.iter() {
+        for (i, batch) in val_loader.iter().enumerate() {
             // Convert batch to tensors
             let (static_tensor, hist_tensor, fut_tensor, target_tensor) =
                 self.batch_to_tensors(batch)?;
@@ -1493,6 +1493,16 @@ impl TFTTrainer {
             }
 
             batch_count += 1;
+
+            // Clear CUDA cache every 10 batches to prevent accumulation (CRITICAL FIX)
+            if i % 10 == 0 && self.device.is_cuda() {
+                // Clear model's attention cache (prevents 2500MB leak during validation)
+                self.model.clear_cache();
+
+                if let Err(e) = Self::sync_cuda_device(&self.device) {
+                    warn!("Failed to sync CUDA during validation batch {}: {}", i, e);
+                }
+            }
         }
 
         // Defensive check: if no validation batches, return zero loss (skip validation)