ollama로 구동하던 부분 제외함

app.py

@@ -27,19 +27,18 @@ def respond(
 
     messages.append({"role": "user", "content": message})
 
-    response = ""
-    return rag.Response(message)
-    # 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
+    response = "" + rag.SearchDocs(message, k=1)
+    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
 
-    #     response += token
-    #     yield response
+        response += token
+        yield response
 
 """
 For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface

rag.py

@@ -34,30 +34,31 @@ model_huggingface = HuggingFaceEmbeddings(model_name = 'jhgan/ko-sroberta-multit
 db = Chroma.from_documents(sourceDocs, model_huggingface)
 
 ## 질의하기
-def searchDocs(question, k=1):
+def SearchDocs(question, k=1):
     results = db.similarity_search_with_relevance_scores(question, k = k)
-    return results
+    merged = ' '.join([sourceDocs[result[0]][0] for result in results])
+    return merged
 
-################
-# 찾은 문서를 프롬프트에 전달하여 LLM으로 답변 생성
-################
-from langchain_community.chat_models import ChatOllama
-llm = ChatOllama(
-    base_url='http://localhost:11434',
-    # model="phi3:medium", # 너무 느려서 mini로 변경
-    model="phi3:mini",
-)
-
-from langchain_core.prompts import ChatPromptTemplate
-
-prompt = ChatPromptTemplate.from_messages([
-    ("system", "Please answer the following question from the document: {document}"),
-    ("user", "{question}"),
-])
-
-# print('-'*50)
-chain = prompt | llm
-def Response(question):
-    searchedDocs = searchDocs(question)
-    mergedDoc = ' '.join(searchedDocs[0][0])
-    return chain.invoke({"question": question, "document": mergedDoc})
+# ################
+# # 찾은 문서를 프롬프트에 전달하여 LLM으로 답변 생성
+# ################
+# from langchain_community.chat_models import ChatOllama
+# llm = ChatOllama(
+#     base_url='http://localhost:11434',
+#     # model="phi3:medium", # 너무 느려서 mini로 변경
+#     model="phi3:mini",
+# )
+
+# from langchain_core.prompts import ChatPromptTemplate
+
+# prompt = ChatPromptTemplate.from_messages([
+#     ("system", "Please answer the following question from the document: {document}"),
+#     ("user", "{question}"),
+# ])
+
+# # print('-'*50)
+# chain = prompt | llm
+# def Response(question):
+#     searchedDocs = SearchDocs(question)
+#     mergedDoc = ' '.join(searchedDocs[0][0])
+#     return chain.invoke({"question": question, "document": mergedDoc})