Update app.py

app.py

@@ -5,14 +5,15 @@ from langchain.document_loaders import PyPDFLoader, Docx2txtLoader, TextLoader
 from langchain.text_splitter import CharacterTextSplitter
 from langchain.vectorstores import Chroma
 from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.llms import HuggingFacePipeline
 from transformers import pipeline
 import gradio as gr
 
-# Workaround for sqlite in HuggingFace Spaces & environments without sqlite3
+# workaround for sqlite in HF spaces
 __import__('pysqlite3')
 sys.modules['sqlite3'] = sys.modules.pop('pysqlite3')
 
-# 📄 Load documents from multiple_docs folder
+# 📄 Load documents from multiple_docs
 docs = []
 for f in os.listdir("multiple_docs"):
     if f.endswith(".pdf"):
@@ -29,11 +30,11 @@ for f in os.listdir("multiple_docs"):
 splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=10)
 docs = splitter.split_documents(docs)
 
-# 🧠 Prepare texts and metadata
+# 🧬 Prepare texts and metadata
 texts = [doc.page_content for doc in docs]
 metadatas = [{"id": i} for i in range(len(texts))]
 
-# 🧬 Embeddings
+# 🧠 Embeddings
 embedding_function = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
 
 # 🗃️ Vectorstore
@@ -44,22 +45,12 @@ vectorstore = Chroma(
 vectorstore.add_texts(texts=texts, metadatas=metadatas)
 vectorstore.persist()
 
-# 🤖 Load free LLM using pipeline
-model_name = "google/flan-t5-large"  # or flan-t5-base if you prefer faster
-generator = pipeline("text2text-generation", model=model_name, device=-1)  # -1 → CPU
+# 🤖 Load free LLM using pipeline + wrap in HuggingFacePipeline
+model_name = "google/flan-t5-large"  # or flan-t5-base for faster
+hf_pipeline = pipeline("text2text-generation", model=model_name, device=-1)  # CPU
+llm = HuggingFacePipeline(pipeline=hf_pipeline)
 
-# 🔗 Wrap pipeline in a callable for LangChain
-class HuggingFaceLLMWrapper:
-    def __init__(self, generator):
-        self.generator = generator
-
-    def __call__(self, prompt, **kwargs):
-        result = self.generator(prompt, max_new_tokens=512, num_return_sequences=1)
-        return result[0]['generated_text']
-
-llm = HuggingFaceLLMWrapper(generator)
-
-# 🔗 Create Conversational QA chain
+# 🔗 Create conversational chain
 chain = ConversationalRetrievalChain.from_llm(
     llm,
     retriever=vectorstore.as_retriever(search_kwargs={'k': 6}),
@@ -67,7 +58,7 @@ chain = ConversationalRetrievalChain.from_llm(
     verbose=False
 )
 
-# 💬 Gradio interface
+# 💬 Gradio UI
 chat_history = []
 
 with gr.Blocks() as demo: