Update app.py

app.py

@@ -1,15 +1,14 @@
 import gradio as gr
 import os
-from langchain_community.vectorstores import FAISS
-from langchain_community.document_loaders import PyPDFLoader
-from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain_community.embeddings import HuggingFaceEmbeddings 
+from langchain.vectorstores import FAISS
+from langchain.document_loaders import PyPDFLoader
+from langchain.embeddings import HuggingFaceEmbeddings
 from langchain.chains import ConversationalRetrievalChain
-from langchain_community.llms import HuggingFaceEndpoint
 from langchain.memory import ConversationBufferMemory
+from langchain.llms import HuggingFaceHub
 
 # Liste der Modelle
-list_llm = ["google/flan-t5-small", "distilbert-base-uncased"]  # Leichtere, CPU-freundliche Modelle
+list_llm = ["google/flan-t5-small", "distilbert-base-uncased"]  # Leichte Modelle für CPU
 list_llm_simple = [os.path.basename(llm) for llm in list_llm]
 
 # PDF-Dokument laden und aufteilen
@@ -19,7 +18,7 @@ def load_doc(list_file_path):
     for loader in loaders:
         pages.extend(loader.load())
     text_splitter = RecursiveCharacterTextSplitter(
-        chunk_size=512,  # Kleinere Chunks für schnellere Verarbeitung
+        chunk_size=512,  # Kleinere Chunks für schnelleres Verarbeiten auf CPU
         chunk_overlap=32
     )
     doc_splits = text_splitter.split_documents(pages)
@@ -33,16 +32,17 @@ def create_db(splits):
 
 # Initialisierung des LLM Chains
 def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db):
-    llm = HuggingFaceEndpoint(
+    llm = HuggingFaceHub(
         repo_id=llm_model,
-        temperature=temperature,
-        max_new_tokens=max_tokens,
-        top_k=top_k
+        model_kwargs={
+            "temperature": temperature,
+            "max_length": max_tokens,
+            "top_k": top_k,
+        }
     )
 
     memory = ConversationBufferMemory(
         memory_key="chat_history",
-        output_key='answer',
         return_messages=True
     )
 
@@ -80,10 +80,8 @@ def format_chat_history(message, chat_history):
 # Chat-Funktion
 def conversation(qa_chain, message, history):
     formatted_chat_history = format_chat_history(message, history)
-    response = qa_chain.invoke({"question": message, "chat_history": formatted_chat_history})
+    response = qa_chain({"question": message, "chat_history": formatted_chat_history})
     response_answer = response["answer"]
-    if "Helpful Answer:" in response_answer:
-        response_answer = response_answer.split("Helpful Answer:")[-1]
     new_history = history + [(message, response_answer)]
     return qa_chain, gr.update(value=""), new_history