Update app.py

app.py

@@ -7,8 +7,8 @@ from threading import Thread
 
 # Loading the tokenizer and model from Hugging Face's model hub.
 if torch.cuda.is_available():
-    tokenizer = AutoTokenizer.from_pretrained("0x7194633/fialka-13B-v4")
-    model = AutoModelForCausalLM.from_pretrained("0x7194633/fialka-13B-v4", load_in_8bit=True, device_map="auto")
+    tokenizer = AutoTokenizer.from_pretrained("sambanovasystems/SambaLingo-Russian-Chat")
+    model = AutoModelForCausalLM.from_pretrained("sambanovasystems/SambaLingo-Russian-Chat", torch_dtype=torch.float16, device_map="auto")
 
 
 # Defining a custom stopping criteria class for the model's text generation.
@@ -57,7 +57,7 @@ def predict(message, history):
 
 # Setting up the Gradio chat interface.
 gr.ChatInterface(predict,
-                 title="Fialka 13B v4",
+                 title="SambaLingo-Russian-Chat",
                  description="Внимание! Все ответы сгенерированы и могут содержать неточную информацию.",
                  examples=['Как приготовить рыбу?', 'Кто президент США?']
                  ).launch()  # Launching the web interface.