Update app.py

app.py

@@ -4,10 +4,10 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
 model_name_or_path = "TheBloke/Unholy-v1-12L-13B-GPTQ"
 # To use a different branch, change revision
 # For example: revision="main"
-model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
+model = AutoModelForCausalLM.from_pretrained( model_name_or_path,
                                              device_map="auto",
                                              trust_remote_code=False,
-                                             revision="main"
+                                             revision="main",device_map="cuda:0"
                                              
                                             )
 
@@ -41,7 +41,7 @@ pipe = pipeline(
     temperature=0.7,
     top_p=0.95,
     top_k=40,
-    repetition_penalty=1.1 , disable_exllama=True
+    repetition_penalty=1.1 
 )
 
 print(pipe(prompt_template)[0]['generated_text'])