Update app.py

app.py

@@ -1,69 +1,126 @@
-import streamlit as st
-import openai
-import os
-from dotenv import load_dotenv
-from sentence_transformers import SentenceTransformer, util
-from langchain.embeddings import OpenAIEmbeddings
-from openai import embeddings_utils
+# import streamlit as st
+# import openai
+# import os
+# from dotenv import load_dotenv
+# from sentence_transformers import SentenceTransformer, util
+# from langchain.embeddings import OpenAIEmbeddings
+# from openai import embeddings_utils
 
-# 定義一個主函数，它將包含我們的 Streamlit 應用邏輯
-def main(api_key):
+# # 定義一個主函数，它將包含我們的 Streamlit 應用邏輯
+# def main(api_key):
 
-    openai.api_key = api_key
+#     openai.api_key = api_key
     
 
-    # 初始化 Sentence Transformer 模型
-    sentence_model = SentenceTransformer('all-MiniLM-L6-v2')
+#     # 初始化 Sentence Transformer 模型
+#     sentence_model = SentenceTransformer('all-MiniLM-L6-v2')
 
-    # 初始化 LangChain OpenAI Embeddings
-    embeddings = OpenAIEmbeddings()
+#     # 初始化 LangChain OpenAI Embeddings
+#     embeddings = OpenAIEmbeddings()
 
-    st.title('智能讀書會問答系統')
+#     st.title('智能讀書會問答系統')
 
-    user_query = st.text_input("請輸入您的問題：", "")
+#     user_query = st.text_input("請輸入您的問題：", "")
 
-    def generate_answers_with_gpt4(question, max_tokens=50, num_responses=3):
-        response = openai.Completion.create(
-            model="gpt-4",  # 根據情況選擇模型
-            prompt=question,
-            temperature=0.7,
-            max_tokens=max_tokens,
-            n=num_responses,
-            stop=None
-        )
-        return [completion['text'].strip() for completion in response['choices']]
+#     def generate_answers_with_gpt4(question, max_tokens=50, num_responses=3):
+#         response = openai.Completion.create(
+#             model="gpt-4",  # 根據情況選擇模型
+#             prompt=question,
+#             temperature=0.7,
+#             max_tokens=max_tokens,
+#             n=num_responses,
+#             stop=None
+#         )
+#         return [completion['text'].strip() for completion in response['choices']]
 
-    def calculate_similarity(answers, user_query):
-        query_embedding = sentence_model.encode(user_query, convert_to_tensor=True)
-        answer_embeddings = sentence_model.encode(answers, convert_to_tensor=True)
-        cosine_scores = util.pytorch_cos_sim(query_embedding, answer_embeddings)
-        return cosine_scores[0].tolist()
+#     def calculate_similarity(answers, user_query):
+#         query_embedding = sentence_model.encode(user_query, convert_to_tensor=True)
+#         answer_embeddings = sentence_model.encode(answers, convert_to_tensor=True)
+#         cosine_scores = util.pytorch_cos_sim(query_embedding, answer_embeddings)
+#         return cosine_scores[0].tolist()
 
-    if st.button('提交'):
-        answers = generate_answers_with_gpt4(user_query)
+#     if st.button('提交'):
+#         answers = generate_answers_with_gpt4(user_query)
         
-        # 將用戶問題轉換成向量 embeddings
-        embedded_query = embeddings.embed_documents([user_query])[0]
+#         # 將用戶問題轉換成向量 embeddings
+#         embedded_query = embeddings.embed_documents([user_query])[0]
         
-        # 計算每個答案與用戶問題的similarity numbers
-        similarities = []
-        for answer in answers:
-            answer_embedding = embeddings.embed_documents([answer])[0]
-            similarity = embeddings_utils.cosine_similarity(embedded_query, answer_embedding)
-            similarities.append(similarity)
+#         # 計算每個答案與用戶問題的similarity numbers
+#         similarities = []
+#         for answer in answers:
+#             answer_embedding = embeddings.embed_documents([answer])[0]
+#             similarity = embeddings_utils.cosine_similarity(embedded_query, answer_embedding)
+#             similarities.append(similarity)
         
-        # 根據相似度分數進行排序 -1~1
-        answers_with_scores = sorted(zip(answers, similarities), key=lambda x: x[1], reverse=True)
+#         # 根據相似度分數進行排序 -1~1
+#         answers_with_scores = sorted(zip(answers, similarities), key=lambda x: x[1], reverse=True)
         
-        for answer, score in answers_with_scores:
-            st.write(f"答案: {answer}\n相似度分數: {score:.2f}")
+#         for answer, score in answers_with_scores:
+#             st.write(f"答案: {answer}\n相似度分數: {score:.2f}")
+
+
+# if __name__ == "__main__":
+#     st.sidebar.title("配置")
+#     api_key = st.sidebar.text_input("請輸入您的 OpenAI API KEY：", type="password")
+    
+#     if api_key:
+#         main(api_key)
+#     else:
+#         st.sidebar.warning("請輸入 OpenAI API KEY 來啟動程式。")
 
+import streamlit as st
+import openai
+from dotenv import load_dotenv
+import os
+# 使用 langchain-community 中的 embeddings 功能
+from langchain_community.embeddings import Embeddings
+
+# 载入环境变量（如 OpenAI API Key）
+load_dotenv()
+
+def main(api_key):
+    openai.api_key = api_key
+
+    st.title('智能阅读会问答系统')
+
+    user_query = st.text_input("请输入您的问题：", "")
+
+    # 初始化 embeddings 对象
+    embeddings = Embeddings()
+
+    def generate_answers_with_gpt_chat(question):
+        response = openai.ChatCompletion.create(
+            model="gpt-3.5-turbo",  # 或根据需要调整为其他模型
+            messages=[{"role": "user", "content": question}]
+        )
+        return [response['choices'][0]['message']['content']]
+
+    def calculate_similarity(answers, user_query):
+        # 获取问题的 embedding
+        query_embedding = embeddings.embed_text(user_query)
+        # 计算每个答案的 embedding 并计算与问题的相似度
+        similarities = [embeddings.similarity(query_embedding, embeddings.embed_text(answer)) for answer in answers]
+        return similarities
+
+    if st.button('提交'):
+        if user_query:
+            answers = generate_answers_with_gpt_chat(user_query)
+            # 计算相似度
+            similarities = calculate_similarity(answers, user_query)
+            # 根据相似度进行排序
+            answers_with_scores = sorted(zip(answers, similarities), key=lambda x: x[1], reverse=True)
+            
+            st.subheader("生成并排序的答案：")
+            for answer, score in answers_with_scores:
+                st.text(f"答案: {answer}\n相似度分数: {score:.2f}")
+        else:
+            st.warning("请输入一个问题。")
 
 if __name__ == "__main__":
     st.sidebar.title("配置")
-    api_key = st.sidebar.text_input("請輸入您的 OpenAI API KEY：", type="password")
+    api_key = st.sidebar.text_input("请输入您的 OpenAI API KEY：", type="password")
     
     if api_key:
         main(api_key)
     else:
-        st.sidebar.warning("請輸入 OpenAI API KEY 來啟動程式。")
+        st.sidebar.warning("请输入 OpenAI API KEY 来启动程序。")