Update

app.py

@@ -1,6 +1,8 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
 
+import rag
+
 """
 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
 """
@@ -26,18 +28,18 @@ 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
 
-    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

Module/rag.py → rag.py

@@ -34,16 +34,9 @@ model_huggingface = HuggingFaceEmbeddings(model_name = 'jhgan/ko-sroberta-multit
 db = Chroma.from_documents(sourceDocs, model_huggingface)
 
 ## 질의하기
-question = '삼성전자의 주요 사업영역은?'
-docs3 = db.similarity_search_with_relevance_scores(question, k = 1) # 2개 하니 너무 느린건가? 답변이 안나오는데..?
-
-# 파일로 저장하고 불러오기
-# db_toFiles = Chroma.from_documents(docs, model_huggingface, persist_directory = './samsumg.db')
-# db_fromfile = Chroma(persist_directory = './samsumg.db',embedding_function=model_huggingface)
-# docs3 = db_fromfile.similarity_search_with_relevance_scores(question,k=3)
-
-joinDoc = ' '.join([doc[0].page_content for doc in docs3])
-print(joinDoc)
+def searchDocs(question, k=1):
+    results = db.similarity_search_with_relevance_scores(question, k = k)
+    return results
 
 ################
 # 찾은 문서를 프롬프트에 전달하여 LLM으로 답변 생성
@@ -62,6 +55,9 @@ prompt = ChatPromptTemplate.from_messages([
     ("user", "{question}"),
 ])
 
-print('-'*50)
+# print('-'*50)
 chain = prompt | llm
-print(chain.invoke({"question": question, "document": joinDoc}))
+def Response(question):
+    searchedDocs = searchDocs(question)
+    mergedDoc = ' '.join(searchedDocs[0][0])
+    return chain.invoke({"question": question, "document": mergedDoc})