fix truncate

app/ner/services/ner.py

@@ -12,7 +12,11 @@ class NER:
         from optimum.onnxruntime import ORTModelForTokenClassification
         from optimum.pipelines import pipeline
 
-        self.tokenizer = AutoTokenizer.from_pretrained(self.model_dir)
+        self.tokenizer = AutoTokenizer.from_pretrained(
+                                                    self.model_dir,
+                                                    truncation=settings.TRUNCATE, 
+                                                    max_length=settings.MAX_LENGTH
+                                                    )
         self.model = ORTModelForTokenClassification.from_pretrained(self.model_dir)
         self.pipeline = pipeline(task=settings.TASK_NAME, 
                                  model=self.model, 
@@ -23,18 +27,7 @@ class NER:
         if self.pipeline is None:
             raise ValueError("Model not loaded. Please call load_model() first.")
         
-        inputs = self.tokenizer(
-            text,
-            truncation=settings.TRUNCATE, 
-            max_length=settings.MAX_LENGTH, 
-            return_tensors="pt"  
-        )
-
-        pred = self.model(**inputs)
-        
-        pred = self.pipeline.decode(pred.logits)
-        
-        # pred = self.pipeline(text)
+        pred = self.pipeline(text)
         
         if entity_tag:
             return self.extract_entities(pred, entity_tag)