diff --git a/llava/model/llava_arch.py b/llava/model/llava_arch.py
index d71650eac..678e27471 100644
--- a/llava/model/llava_arch.py
+++ b/llava/model/llava_arch.py
@@ -257,7 +257,12 @@ def prepare_inputs_labels_for_multimodal(
                 cur_new_input_embeds.append(cur_input_embeds_no_im[i])
                 cur_new_labels.append(cur_labels_noim[i])
                 if i < num_images:
-                    cur_image_features = image_features[cur_image_idx]
+                    try:
+                        cur_image_features = image_features[cur_image_idx]
+                    except Exception as e:
+                            print(f'Tracking input_ids START.....{e}')
+                            print(input_ids)
+                            print('Tracking input_ids END.....')
                     cur_image_idx += 1
                     cur_new_input_embeds.append(cur_image_features)
                     cur_new_labels.append(torch.full((cur_image_features.shape[0],), IGNORE_INDEX, device=cur_labels.device, dtype=cur_labels.dtype))
diff --git a/llava/train/train.py b/llava/train/train.py
index d8442f265..1a3bf122a 100644
--- a/llava/train/train.py
+++ b/llava/train/train.py
@@ -750,6 +750,9 @@ def expand2square(pil_img, background_color):
             # image does not exist in the data, but the model is multimodal
             crop_size = self.data_args.image_processor.crop_size
             data_dict['image'] = torch.zeros(3, crop_size['height'], crop_size['width'])
+        print(sources)
+        print(data_dict)
+        print('Tracking source and data_dict above.....')
         return data_dict