Update app.py

app.py

@@ -10,7 +10,8 @@ from transformers import (
     AutoConfig,
     AutoModelForCausalLM,
     AutoTokenizer,
-    GenerationConfig
+    GenerationConfig,
+    pipeline
 )
 import boto3
 import torch
@@ -57,6 +58,13 @@ class S3ModelLoader:
             config = AutoConfig.from_pretrained(f"s3://{self.bucket_name}/{model_name}")
             model = AutoModelForCausalLM.from_pretrained(f"s3://{self.bucket_name}/{model_name}", config=config)
             tokenizer = AutoTokenizer.from_pretrained(f"s3://{self.bucket_name}/{model_name}")
+
+            # Asegurarse de que el `eos_token_id` y `pad_token_id` estén definidos
+            if model.config.eos_token_id is None:
+                model.config.eos_token_id = tokenizer.eos_token_id
+            if model.config.pad_token_id is None:
+                model.config.pad_token_id = tokenizer.pad_token_id
+
             return model, tokenizer
         except Exception:
             return None, None
@@ -72,9 +80,16 @@ class S3ModelLoader:
 
     async def download_and_save_model_from_huggingface(self, model_name):
         try:
-            # Eliminado tqdm para descargar sin mostrar progreso
+            # Descarga del modelo sin tqdm
             model = AutoModelForCausalLM.from_pretrained(model_name, use_auth_token=HUGGINGFACE_HUB_TOKEN)
             tokenizer = AutoTokenizer.from_pretrained(model_name, use_auth_token=HUGGINGFACE_HUB_TOKEN)
+
+            # Asegurarse de que el `eos_token_id` y `pad_token_id` estén definidos
+            if model.config.eos_token_id is None:
+                model.config.eos_token_id = tokenizer.eos_token_id
+            if model.config.pad_token_id is None:
+                model.config.pad_token_id = tokenizer.pad_token_id
+
             self.upload_model_to_s3(model_name, model, tokenizer)
             return model, tokenizer
         except Exception as e: