Update handler.py

handler.py

@@ -52,8 +52,8 @@ class EndpointHandler():
         else:
             output = self.model.generate(input_ids=input_ids, attention_mask=attention_mask)
         # Postprocess
-        LOGGER.info(output)
-        print(output)
-        prediction = self.tokenizer.decode(output[0])
+
+        prediction = self.tokenizer.decode(output[0], skip_special_tokens=True)
         LOGGER.info(f"Generated text: {prediction}")
-        return {"generated_text": prediction}
+        
+        return [{"generated_text": prediction}]