Update app.py

app.py

@@ -91,11 +91,29 @@ def image_generation(query):
 def doc_question_answering(query, file_path):
     with open(file_path, 'r') as f:
         file_content = f.read()
+
+    # Split the document into smaller chunks
     text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
     docs = text_splitter.create_documents([file_content])
-    embeddings = OpenAIEmbeddings()
+
+    # Create embeddings using the groq model
+    embeddings = OpenAIEmbeddings()  # If you're using a custom embeddings model, replace this line with the corresponding embeddings model for groq
+
+    # Set up the Chroma database for document retrieval
     db = Chroma.from_documents(docs, embeddings, persist_directory=".chroma_db")
-    qa = RetrievalQA.from_chain_type(llm=OpenAI(), chain_type="stuff", retriever=db.as_retriever())
+
+    # Create a custom function to use groq for the question-answering step
+    def groq_llm(query):
+        response = client.chat.completions.create(
+            model=MODEL,
+            messages=[{"role": "user", "content": query}]
+        )
+        return response.choices[0].message.content
+
+    # Set up the RetrievalQA chain using the custom groq LLM function
+    qa = RetrievalQA.from_chain_type(llm=groq_llm, chain_type="stuff", retriever=db.as_retriever())
+
+    # Run the QA process with the groq model
     return qa.run(query)
 
 # Function to handle different input types and choose the right tool