add

app.py

@@ -20,7 +20,7 @@ def load_model(model_name="deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"):
 def main():
     st.set_page_config(page_title="DeepSeek-R1 Chatbot", page_icon="🤖")
     st.title("DeepSeek-R1 기반 대화형 챗봇")
-    st.write("DeepSeek-R1-Distill-Qwen-1.5B 모델을 활용한 한국어 대화 테스트용 데모입니다.")
+    st.write("DeepSeek-R1-Distill-Qwen-1.5B 모델을 사용한 대화 테스트용 데모입니다.")
     
     # 세션 스테이트 초기화
     if "chat_history_ids" not in st.session_state:
@@ -44,7 +44,7 @@ def main():
                 st.write(bot_text)
     
     # 채팅 입력창
-    user_input = st.chat_input("메시지를 입력하세요...")
+    user_input = st.chat_input("영어로 메시지를 입력하세요...")
     
     if user_input:
         # 사용자 메시지 표시

app2.py

@@ -0,0 +1,56 @@
+import torch
+from transformers import pipeline
+import gradio as gr
+
+# 모델 로드
+def load_model(model_name="deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"):
+    pipe = pipeline(
+        "text-generation",
+        model=model_name,
+        device_map="auto",
+        torch_dtype=torch.float16,
+        trust_remote_code=True,
+        truncation=True,
+        max_new_tokens=2048
+    )
+    return pipe
+
+# 챗봇 응답 생성 함수
+def chat(message, history):
+    pipe = load_model()
+    prompt = f"Human: {message}\n\nAssistant:"
+    
+    response = pipe(
+        prompt,
+        max_new_tokens=2048,
+        temperature=0.7,
+        do_sample=True,
+        truncation=True,
+        pad_token_id=50256
+    )
+    
+    # 응답 추출 및 처리
+    try:
+        bot_text = response[0]["generated_text"]
+        bot_text = bot_text.split("Assistant:")[-1].strip()
+        if "</think>" in bot_text:
+            bot_text = bot_text.split("</think>")[-1].strip()
+    except:
+        bot_text = "Sorry, there was a problem generating the response."
+    
+    return bot_text
+
+# Gradio 인터페이스 설정
+demo = gr.ChatInterface(
+    chat,
+    chatbot=gr.Chatbot(height=600),
+    textbox=gr.Textbox(placeholder="Enter your message...", container=False, scale=7),
+    title="DeepSeek-R1 Chatbot",
+    description="DeepSeek-R1-Distill-Qwen-1.5B 모델을 사용한 대화 테스트용 데모입니다.",
+    examples=["Hello", "Who are you?", "What can you do?"],
+    theme=gr.themes.Soft()
+)
+
+# 서버 실행
+if __name__ == "__main__":
+    demo.launch(share=True, server_name="0.0.0.0") 

requirements.txt

@@ -1,4 +1,5 @@
 streamlit
 torch
 transformers>=4.35.0
-accelerate>=0.26.0
+accelerate
+gradio>=4.0.0