Implement gc.collect() to free memory (#10129)

Signed-off-by: Glenn Jocher <glenn.jocher@ultralytics.com>
Co-authored-by: UltralyticsAssistant <web@ultralytics.com>

ultralytics/engine/trainer.py

@@ -6,6 +6,7 @@ Usage:
     $ yolo mode=train model=yolov8n.pt data=coco128.yaml imgsz=640 epochs=100 batch=16
 """
 
+import gc
 import math
 import os
 import subprocess
@@ -437,6 +438,7 @@ class BaseTrainer:
                 self.scheduler.last_epoch = self.epoch  # do not move
                 self.stop |= epoch >= self.epochs  # stop if exceeded epochs
             self.run_callbacks("on_fit_epoch_end")
+            gc.collect()
             torch.cuda.empty_cache()  # clear GPU memory at end of epoch, may help reduce CUDA out of memory errors
 
             # Early Stopping
@@ -458,6 +460,7 @@ class BaseTrainer:
             if self.args.plots:
                 self.plot_metrics()
             self.run_callbacks("on_train_end")
+        gc.collect()
         torch.cuda.empty_cache()
         self.run_callbacks("teardown")