diff --git a/exps/muti_label/dataset.py b/exps/muti_label/dataset.py
index e23db0a..a73f876 100644
--- a/exps/muti_label/dataset.py
+++ b/exps/muti_label/dataset.py
@@ -22,8 +22,11 @@ def parse_data_info(self, raw_data_info):
         # print(raw_img_info)
         # print([cur_cates[ann['category_id']-1] for ann in raw_ann_info])
         # to one-hot
-        category_ids = torch.unique(torch.tensor([ann['category_id'] for ann in raw_ann_info])) - 1
-        cate_one_hot = torch.eye(len(cur_cates))[category_ids].sum(dim=0)
+        if len(raw_ann_info) == 0:
+            cate_one_hot = torch.zeros(len(cur_cates))
+        else:
+            category_ids = torch.unique(torch.tensor([ann['category_id'] for ann in raw_ann_info])) - 1
+            cate_one_hot = torch.eye(len(cur_cates))[category_ids].sum(dim=0)
 
         return {
             "img_path": os.path.join(self.data_root, raw_img_info['file_name']),
@@ -45,9 +48,9 @@ def load_data_list(self) -> list[dict]:
             ann_ids = self.lvis.get_ann_ids(img_ids=[img_id])
             raw_ann_info = self.lvis.load_anns(ann_ids)
 
-            if len(raw_ann_info) == 0:
-                # print(f"Image {img_id} has no annotations, skipped.")
-                continue
+            # if len(raw_ann_info) == 0:
+            #     # print(f"Image {img_id} has no annotations, skipped.")
+            #     continue
 
             parsed_data_info = self.parse_data_info({
                 'raw_ann_info':
diff --git a/oadp/dp/roi_heads.py b/oadp/dp/roi_heads.py
index 3c52b5b..37a62a8 100644
--- a/oadp/dp/roi_heads.py
+++ b/oadp/dp/roi_heads.py
@@ -191,8 +191,8 @@ def forward(
             img_path = img_meta['img_path']
             if img_path not in self.ram_pred_result:
                 print(f"Image path {img_path} not found in RAM prediction results")
-                ram_cls_score.append(np.ones((1, cls_score.shape[1])))
-                raise ValueError(f"Image path {img_path} not found in RAM prediction results")
+                ram_cls_score.append(np.ones((cls_score.shape[1] - 1, )))
+                # raise ValueError(f"Image path {img_path} not found in RAM prediction results")
             else:
                 ram_cls_score.append(self.ram_pred_result[img_path])
         ram_cls_score = np.concatenate(ram_cls_score, axis=0)
@@ -203,8 +203,8 @@ def forward(
         ram_cls_score = repeat(ram_cls_score, 'c -> b c', b=num_box).sigmoid()
         ram_cls_score_with_bg = torch.cat([ram_cls_score, background_score], dim=1).to(cls_score.device)
 
-        return cls_score * ram_cls_score_with_bg
-    
+        return (cls_score.softmax(-1) * ram_cls_score_with_bg).log()
+
     
 @MODELS.register_module()
 class RAMEnsembleOADPRoIHead(OADPRoIHead):