Update app.py

app.py

@@ -4,12 +4,13 @@ import hashlib
 import pickle
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_community.document_loaders import PyPDFLoader
-from langchain_openai import OpenAIEmbeddings, ChatOpenAI
+from langchain_community.embeddings import HuggingFaceEmbeddings
 from langchain_community.vectorstores import FAISS
 from langchain.chains import RetrievalQA
+from langchain_groq import ChatGroq
 
-# Load/OpenAI key from Hugging Face Secrets
-OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
+# Load Groq API key
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
 
 # Directory to cache vectorstores
 CACHE_DIR = "vector_cache"
@@ -28,7 +29,7 @@ def build_vectorstore(pdf_path: str):
     loader = PyPDFLoader(pdf_path)
     documents = loader.load()
 
-    # Chunking strategy (important for assignment accuracy)
+    # Chunking strategy
     text_splitter = RecursiveCharacterTextSplitter(
         chunk_size=1000,
         chunk_overlap=200,
@@ -36,7 +37,7 @@ def build_vectorstore(pdf_path: str):
     )
     chunks = text_splitter.split_documents(documents)
 
-    embeddings = OpenAIEmbeddings(openai_api_key=OPENAI_API_KEY)
+    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
     vectorstore = FAISS.from_documents(chunks, embeddings)
     return vectorstore
 
@@ -67,12 +68,14 @@ def rag_bot(question: str, pdf_path: str):
         vectorstore = get_vectorstore(pdf_path)
         retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
 
+        # Use Groq LLM
+        llm = ChatGroq(
+            groq_api_key=GROQ_API_KEY,
+            model_name="mixtral-8x7b-32768",  # or "llama2-70b-4096"
+        )
+
         qa = RetrievalQA.from_chain_type(
-            llm=ChatOpenAI(
-                model="gpt-3.5-turbo",
-                temperature=0,
-                openai_api_key=OPENAI_API_KEY
-            ),
+            llm=llm,
             chain_type="stuff",
             retriever=retriever,
         )
@@ -84,7 +87,7 @@ def rag_bot(question: str, pdf_path: str):
 
 # ------------------ Gradio UI ------------------
 with gr.Blocks() as demo:
-    gr.Markdown("## 📖 RAG Q&A Bot – Upload a PDF and Ask Questions")
+    gr.Markdown("## 📖 RAG Q&A Bot – Powered by Groq + HuggingFace Embeddings")
 
     with gr.Row():
         pdf_file = gr.File(label="Upload PDF", type="filepath", file_types=[".pdf"])