Update

Signed-off-by: Jael Gu <mengjia.gu@zilliz.com>

README.md

@@ -48,7 +48,7 @@ Create the operator via the following factory method:
 The model name in string, defaults to None.
 If None, the operator will be initialized without specified model.
 
-Supported model names:
+Please note only supported models are tested by us:
 
 <details><summary>Albert</summary>
 
@@ -307,6 +307,12 @@ Supported model names:
 The path to local checkpoint, defaults to None.
 If None, the operator will download and load pretrained model by `model_name` from Huggingface transformers.
 
+<br />
+
+***device***: *str*
+
+The device in string, defaults to None. If None, it will enable "cuda" automatically when cuda is available.
+
 <br /> 
 
 ***tokenizer***: *object*

auto_transformers.py

@@ -37,14 +37,33 @@ warnings.filterwarnings('ignore')
 os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
 t_logging.set_verbosity_error()
 
+def create_model(model_name, checkpoint_path, device):
+    model = AutoModel.from_pretrained(model_name).to(device)
+    if hasattr(model, 'pooler') and model.pooler:
+        model.pooler = None
+    if checkpoint_path:
+        try:
+            state_dict = torch.load(checkpoint_path, map_location=device)
+            model.load_state_dict(state_dict)
+        except Exception:
+            log.error(f'Fail to load weights from {checkpoint_path}')
+    model.eval()
+    return model
 
 # @accelerate
 class Model:
-    def __init__(self, model):
-        self.model = model
+    def __init__(self, model_name, checkpoint_path, device):
+        self.device = device
+        self.model = create_model(model_name, checkpoint_path, device)
 
     def __call__(self, *args, **kwargs):
-        outs = self.model(*args, **kwargs, return_dict=True)
+        new_args = []
+        for x in args:
+            new_args.append(x.to(self.device))
+        new_kwargs = {}
+        for k, v in kwargs.items():
+            new_kwargs[k] = v.to(self.device)
+        outs = self.model(*new_args, **new_kwargs, return_dict=True)
         return outs['last_hidden_state']
 
 
@@ -75,17 +94,13 @@ class AutoTransformers(NNOperator):
         self.checkpoint_path = checkpoint_path
 
         if self.model_name:
-            model_list = self.supported_model_names()
-            assert model_name in model_list, f"Invalid model name: {model_name}. Supported model names: {model_list}"
-            self.model = Model(self._model)
-            if tokenizer is None:
-                try:
-                    self.tokenizer = AutoTokenizer.from_pretrained(model_name)
-                except Exception as e:
-                    log.error(f'Fail to load default tokenizer by name: {self.model_name}')
-                    raise e
-            else:
+            # model_list = self.supported_model_names()
+            # assert model_name in model_list, f"Invalid model name: {model_name}. Supported model names: {model_list}"
+            self.model = Model(model_name=self.model_name, checkpoint_path=self.checkpoint_path, device=self.device)
+            if tokenizer:
                 self.tokenizer = tokenizer
+            else:
+                self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
             if not self.tokenizer.pad_token:
                 self.tokenizer.pad_token = '[PAD]'
         else:
@@ -98,7 +113,7 @@ class AutoTransformers(NNOperator):
         else:
             txt = data
         try:
-            inputs = self.tokenizer(txt, padding=True, truncation=True, return_tensors="pt").to(self.device)
+            inputs = self.tokenizer(txt, padding=True, truncation=True, return_tensors='pt')
         except Exception as e:
             log.error(f'Fail to tokenize inputs: {e}')
             raise e
@@ -116,17 +131,7 @@ class AutoTransformers(NNOperator):
 
     @property
     def _model(self):
-        model = AutoModel.from_pretrained(self.model_name).to(self.device)
-        if hasattr(model, 'pooler') and model.pooler:
-            model.pooler = None
-        if self.checkpoint_path:
-            try:
-                state_dict = torch.load(self.checkpoint_path, map_location=self.device)
-                model.load_state_dict(state_dict)
-            except Exception:
-                log.error(f'Fail to load weights from {self.checkpoint_path}')
-        model.eval()
-        return model
+        return self.model.model
 
     def save_model(self, model_type: str = 'pytorch', output_file: str = 'default'):
         if output_file == 'default':
@@ -160,6 +165,7 @@ class AutoTransformers(NNOperator):
         elif model_type == 'onnx':
             from transformers.onnx.features import FeaturesManager
             from transformers.onnx import export
+            self._model = self._model.to('cpu')
             model_kind, model_onnx_config = FeaturesManager.check_supported_model_or_raise(
                 self._model, feature='default')
             onnx_config = model_onnx_config(self._model.config)