transformerModel

app.py

@@ -160,11 +160,12 @@ if __name__ == "__main__":
 
     # you can use HFApiModel from smolagents if you don't mind paying more money.
     model = TransformersModel(
-        # model_id="Qwen/Qwen2.5-Coder-14B-Instruct",
-        model_id="meta-llama/Llama-3.2-3B-Instruct",
+        model_id="Qwen/Qwen2.5-Coder-14B-Instruct",
+        # model_id="meta-llama/Llama-3.2-3B-Instruct",
         device_map="cuda"
+        ,max_new_tokens=5000
 
-    )
+    ).bfloat16().cuda()
     # model = HfApiModel(
         # model_id="meta-llama/Meta-Llama-3.1-8B-Instruct",
         # token = <YOUR-HF-TOKEN-HERE>