Support loading trained weights path.

Signed-off-by: wxywb <xy.wang@zilliz.com>

clip.py

@@ -38,13 +38,11 @@ os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
 t_logging.set_verbosity_error()
 
 def create_model(model_name, modality, checkpoint_path, device):
-    hf_clip_model = CLIPModel.from_pretrained(model_name)
-    if checkpoint_path:
-        try:
-            state_dict = torch.load(checkpoint_path, map_location=device)
-            hf_clip_model.load_state_dict(state_dict)
-        except Exception as e:
-            log.error(f"Fail to load state dict from {checkpoint_path}: {e}")
+    if checkpoint_path is None:
+        hf_clip_model = CLIPModel.from_pretrained(model_name)
+    else:
+        hf_clip_config = CLIPModel.from_config(model_name)
+        hf_clip_model = CLIPModel.from_pretrained(checkpoint_path, config=hf_clip_config)
     hf_clip_model.to(device)
     hf_clip_model.eval()
 
@@ -101,7 +99,10 @@ class Clip(NNOperator):
         self.modality = modality
         self.device = device
         self.checkpoint_path = checkpoint_path
-        real_name = self._configs()[model_name]
+        config = self._configs()
+        real_name = model_name
+        if model_name in config:
+            real_name = config[model_name]
 
         self.model = Model(real_name, modality, checkpoint_path, device)
         self.tokenizer = CLIPTokenizer.from_pretrained(real_name)