Update app.py

app.py

@@ -17,16 +17,24 @@ For more information on `huggingface_hub` Inference API support, please check th
 client = InferenceClient("meta-llama/Llama-3.1-8B-Instruct", token=os.getenv('deepseekv2'))
 
 
-@spaces.GPU(duration=13)
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    model,
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
+def choose_model(model_name):
+    if model_name == "Qwen1.5":
+        client = InferenceClient("deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B", token=os.getenv('deepseekv2'))
+        
+    elif model_name == "Llama3-8b-Instruct":    
+        client = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct", token=os.getenv('deepseekv2'))
+
+    elif model_name == "Llama3.1-8b-Instruct"
+        client = InferenceClient("meta-llama/Llama-3.1-8B-Instruct", token=os.getenv('deepseekv2'))
+    
+    return 
+    
+
+@spaces.GPU(duration=1)
+def respond(message, history: list[tuple[str, str]], model, system_message, max_tokens, temperature, top_p,):
+
+    print(model)
+    choose_model(model)
     messages = [{"role": "system", "content": system_message}]
 
     for val in history:
@@ -39,13 +47,7 @@ def respond(
 
     response = ""
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
+    for message in client.chat_completion(messages, max_tokens=max_tokens, stream=True, temperature=temperature, top_p=top_p):
         token = message.choices[0].delta.content
 
         response += token
@@ -66,13 +68,7 @@ demo = gr.ChatInterface(
         gr.Textbox(value="You are a friendly and helpful Chatbot, be concise and straight to the point, avoid excessive reasoning.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
     ],
 )