app.py

app.py

@@ -0,0 +1,75 @@
+from langchain.document_loaders import DirectoryLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings.openai import OpenAIEmbeddings
+from langchain.vectorstores import Chroma
+from langchain.chat_models import ChatOpenAI
+from langchain.retrievers.multi_query import MultiQueryRetriever
+import dotenv
+from langchain.indexes import VectorstoreIndexCreator
+from langchain.chains.question_answering import load_qa_chain
+from langchain.llms import OpenAI
+from langchain.prompts import PromptTemplate
+from langchain.chat_models import ChatOpenAI
+from langchain.schema import AIMessage, HumanMessage, SystemMessage
+import gradio as gr
+
+dotenv.load_dotenv()
+
+
+system_message = """You are the helpful assistant representing the company ecredit. 
+You answers should be in Greek. 
+If you don't know the answer, just say that you don't know, don't try to make up an answer.
+Always finish your answer with "για περισσότερες πληροφορίες καλέστε στο: XXXXXXXXXXX.".
+"""
+
+prompt_template = """Use the following pieces of context to answer the question at the end. 
+If you don't know the answer, just say that you don't know, don't try to make up an answer.
+Only answer questions that are related to the context. If it's not in the context say "Δεν γνωρίζω".
+
+Context:
+{context}
+
+Question: {question}
+Answer in Greek:
+"""
+PROMPT = PromptTemplate(
+    template=prompt_template, input_variables=["context", "question"]
+)
+
+loader = DirectoryLoader("./documents", glob="**/*.pdf", show_progress=True)
+docs = loader.load()
+text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0)
+texts = text_splitter.split_documents(docs)
+
+embeddings = OpenAIEmbeddings()
+docsearch = Chroma.from_documents(texts, embeddings).as_retriever()
+chat = ChatOpenAI(temperature=0.1)
+
+
+with gr.Blocks() as demo:
+    chatbot = gr.Chatbot()
+    msg = gr.Textbox()
+    clear = gr.ClearButton([msg, chatbot])
+
+    def respond(message, chat_history):
+        messages = [
+            SystemMessage(content=system_message),
+        ]
+
+        result_docs = docsearch.get_relevant_documents(message)
+        human_message = None
+        human_message = HumanMessage(
+            content=PROMPT.format(context=result_docs[:3], question=message)
+        )
+        messages.append(human_message)
+
+        result = chat(messages)
+        bot_message = result.content
+        chat_history.append((message, bot_message))
+        return "", chat_history
+
+    msg.submit(respond, [msg, chatbot], [msg, chatbot])
+
+
+if __name__ == "__main__":
+    demo.launch()