Update app.py

app.py

@@ -1,64 +1,66 @@
+import json
 import gradio as gr
-from huggingface_hub import InferenceClient
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.metrics.pairwise import cosine_similarity
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+file_path = "InstructData.jsonl"
 
+# 1. 대화 데이터 불러오기 (모든 human-gpt 쌍 추출)
+def load_conversations(file_path):
+    conversations = []
+    with open(file_path, 'r', encoding='utf-8') as f:
+        for line in f:
+            data = json.loads(line)
+            conv = data.get('conversations', [])
+            prev = None
+            for c in conv:
+                if c.get('from') == 'human':
+                    prev = c['value']
+                elif c.get('from') == 'gpt' and prev:
+                    conversations.append((prev, c['value']))
+                    prev = None
+    return conversations
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# 2. 한국어 불용어 설정 + TF-IDF 벡터라이저 준비
+def build_vectorizer(questions):
+    korean_stopwords = ['그리고', '그러나', '하지만', '또한', '이런', '저런', '그런', '있는', '없는', '것', '수', '때문에', '해서']
+    vectorizer = TfidfVectorizer(stop_words=korean_stopwords)
+    tfidf_matrix = vectorizer.fit_transform(questions)
+    return vectorizer, tfidf_matrix
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+# 3. 유사도 기반 Top-k 응답 찾기
+def find_top_responses(user_input, conversations, vectorizer, tfidf_matrix, top_k=3):
+    user_vec = vectorizer.transform([user_input])
+    sims = cosine_similarity(user_vec, tfidf_matrix)[0]
+    top_indices = sims.argsort()[-top_k:][::-1]
+    results = ""
+    for i, idx in enumerate(top_indices, 1):
+        answer = conversations[idx][1]
+        score = sims[idx]
+        results += f"{i}. {answer} (유사도: {score:.2f})\n\n"
+    return results.strip()
 
-    messages.append({"role": "user", "content": message})
+# 초기화
+convs = load_conversations(file_path)
+questions = [q for q, _ in convs]
+vectorizer, tfidf_matrix = build_vectorizer(questions)
 
-    response = ""
+# Gradio 인터페이스 함수
+def chatbot_interface(user_input):
+    if user_input.strip().lower() == "종료":
+        return "ChatBot: 안녕~ 또 놀러와!"
+    return find_top_responses(user_input, convs, vectorizer, tfidf_matrix)
 
-    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
+# Gradio UI 구성
+with gr.Blocks() as demo:
+    gr.Markdown("##  ChatBot (TF-IDF 기반 Top-3 답변 추천)")
+    with gr.Row():
+        with gr.Column():
+            input_box = gr.Textbox(label="질문을 입력해줘!", placeholder="예: 챗봇 어떻게 만들어요?")
+            submit_btn = gr.Button("Toki에게 물어보기 🧠")
+        with gr.Column():
+            output_box = gr.Textbox(label="Toki의 추천 답변 Top-3", lines=10)
 
-        response += token
-        yield response
+    submit_btn.click(fn=chatbot_interface, inputs=input_box, outputs=output_box)
 
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", 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)",
-        ),
-    ],
-)
-
-
-if __name__ == "__main__":
-    demo.launch()
+demo.launch()