Update app.py

app.py

@@ -1,64 +1,49 @@
-import gradio as gr
-from huggingface_hub import InferenceClient
+import pandas as pd
+import faiss
+import numpy as np
+from sentence_transformers import SentenceTransformer
+
+# Загружаем датасет с вопросами и ответами
+train_df = pd.read_csv("data/ostap_train.csv")
+questions = train_df["question"].tolist()
+answers = train_df["answer"].tolist()
 
-"""
-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")
+# Инициализируем эмбеддинг-модель
+model_bge = SentenceTransformer("fitlemon/bge-m3-ru-ostap")
 
+# Вычисляем эмбеддинги для всех вопросов
+answer_embeddings = model_bge.encode(answers, convert_to_numpy=True)
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# Создаём FAISS-индекс на базе вопросов, но в качестве метаданных нужно положить еще ответы
+index = faiss.IndexIDMap(faiss.IndexFlatIP(answer_embeddings.shape[1]))
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+# Добавляем вопросы в индекс
+index.add_with_ids(answer_embeddings, np.arange(len(answers)))
 
-    messages.append({"role": "user", "content": message})
 
-    response = ""
+import gradio as gr
+import time
 
-    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
+with gr.Blocks() as app:
+    chatbot = gr.Chatbot(type="messages")
+    msg = gr.Textbox(
+        label="Напиши свой вопрос Остапу Бендеру здесь...",
+        placeholder="Привет, Остап!",
+    )
+    clear = gr.ClearButton([msg, chatbot])
 
-        response += token
-        yield response
+    def respond(message, chat_history):
 
+        query_emb = model_bge.encode([message], convert_to_numpy=True)
 
-"""
-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)",
-        ),
-    ],
-)
+        _, idx = index.search(query_emb, 1)
+        bot_message = answers[idx[0][0]]
+        chat_history.append({"role": "user", "content": message})
+        chat_history.append({"role": "assistant", "content": bot_message})
+        time.sleep(2)
+        return "", chat_history
 
+    msg.submit(respond, [msg, chatbot], [msg, chatbot])
 
 if __name__ == "__main__":
-    demo.launch()
+    app.launch()