Update handler.py

handler.py

@@ -3,9 +3,9 @@ from transformers import AutoModelForSeq2SeqLM, AutoTokenizer, pipeline
 import torch
 
 class EndpointHandler:
-    def __init__(self, path=""):
+    def __init__(self, path="atwine/translation-en-lug-v6"):
         # Load model and processor from path
-        self.model = AutoModelForSeq2SeqLM.from_pretrained(path, device_map="auto", load_in_8bit=True)
+        self.model = AutoModelForSeq2SeqLM.from_pretrained(path)
         self.tokenizer = AutoTokenizer.from_pretrained(path)
 
     def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
@@ -24,11 +24,15 @@ class EndpointHandler:
         # Modify parameters to increase max_length
         if parameters is None:
             parameters = {}
-        parameters['max_length'] = 1012,  # Set your desired max_length here
-        parameters['min_length'] = 250,
-        parameters['length_penalty'] = 2.0,
-        parameters['num_beams'] = 4,
-        parameters['early_stopping'] = True,
+        parameters['max_length'] = 512  # Set your desired max_length here
+        parameters['min_length'] = 100
+        parameters['length_penalty'] = 2.0
+        parameters['num_beams'] = 10
+        parameters['early_stopping'] = True
+        parameters['temperature'] = 0.0
+        parameters['top_k'] = 15
+        parameters['top_p'] = 0.8
+        
         
 
         # Generate output