Update services/model_service.py

services/model_service.py

@@ -28,7 +28,10 @@ class ModelService:
             logger.info("Loading models...")
 
             # Load tokenizer
-            self.tokenizer = AutoTokenizer.from_pretrained(settings.MODEL_NAME)
+            #self.tokenizer = AutoTokenizer.from_pretrained(settings.MODEL_NAME)
+            self.tokenizer = AutoTokenizer.from_pretrained(settings.MODEL_NAME, use_fast=False)
+            self.tokenizer.pad_token = self.tokenizer.eos_token
+
             logger.info(f"Tokenizer for {settings.MODEL_NAME} loaded successfully.")
 
             # Load language model
@@ -45,7 +48,9 @@ class ModelService:
             logger.info(f"Model {settings.MODEL_NAME} loaded successfully on {quantization_device}.")
 
             # Load sentence embedder
-            self.embedder = SentenceTransformer(settings.EMBEDDER_MODEL)
+            self.embedder = SentenceTransformer(settings.EMBEDDER_MODEL, device='cuda' if torch.cuda.is_available() else 'cpu')
+
+            #self.embedder = SentenceTransformer(settings.EMBEDDER_MODEL)
             logger.info(f"Embedder {settings.EMBEDDER_MODEL} loaded successfully.")
 
         except Exception as e: