Update app.py

app.py

@@ -1,5 +1,3 @@
-
-import time
 import gradio as gr
 from huggingface_hub import InferenceClient
 from optimum.intel import OVModelForCausalLM
@@ -14,40 +12,20 @@ tokenizer = AutoTokenizer.from_pretrained(model_id)
 pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
 
 def respond(message, history):
-    start_time = time.time()  # 記錄開始時間
-
     # 將當前訊息與歷史訊息合併
     input_text = message if not history else history[-1]["content"] + " " + message
     input_text = message
-    # 設定生成參數
-    max_length = 1024  # 增加最大生成長度
-    output_text = ""
-    stop_token = "<|endoftext|>"  # 可選：結束標記
-    while True:
-        response = pipe(input_text, max_length=max_length, truncation=False, num_return_sequences=1)
-        reply = response[0]['generated_text']
-        output_text += reply
-        
-        # 檢測是否包含結束標記，或者生成結束
-        if stop_token in reply or len(output_text) >= max_length:
-            output_text = output_text.split(stop_token)[0]  # 去掉結束標記以後的部分
-            break
-        
-        # 更新輸入文字繼續生成
-        input_text = reply
-
-    end_time = time.time()  # 記錄結束時間
-    duration = end_time - start_time  # 計算耗時
-
-    # 輸出耗時到控制台
+    # 獲取模型的回應
+    response = pipe(input_text, max_length=1024, truncation=True, num_return_sequences=1)
+    reply = response[0]['generated_text']
+    
+    # 返回新的消息格式
     print(f"Message: {message}")
-    print(f"Reply: {output_text}")
-    print(f"Time taken for response: {duration:.2f} seconds")
-
-    return output_text
-
+    print(f"Reply: {reply}")
+    return reply
+    
 # 設定 Gradio 的聊天界面
-demo = gr.ChatInterface(fn=respond, title="Phi-3.5-mini-instruct-openvino", description="Phi-3.5-mini-instruct-openvino", type='messages')
+demo = gr.ChatInterface(fn=respond, title="Chat with Phi-3.5-mini-instruct-openvino", description="與Phi-3.5-mini-instruct-openvino 聊天！", type='messages')
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()