Merge pull request #726 from sneha-4-22/main

Quest_Submission[OS InsightX]

Quest/OS_InsightX/README.md

@@ -0,0 +1,17 @@
+## APP PREVIEW
+
+
+
+https://github.com/sneha-4-22/RAG-OS-qa/assets/112711068/04735392-d9e1-46ea-951b-17ab7a662f58
+
+
+## Model Used
+
+The model used in this project is [Industry-BERT for Insurance](https://huggingface.co/llmware/industry-bert-insurance-v0.1) provided by Hugging Face. It was employed for the operating system question-answering task using the RAG (Retrieval-Augmented Generation) framework.
+
+
+
+
+![image](https://github.com/sneha-4-22/RAG-OS-qa/assets/112711068/749d4468-ca67-4c54-9777-8f3673104151)
+
+

Quest/OS_InsightX/app.py

@@ -0,0 +1,88 @@
+import os
+from dotenv import load_dotenv
+from langchain_community.llms import HuggingFaceHub
+from langchain.chains import RetrievalQA
+from langchain.prompts import PromptTemplate
+from langchain_community.vectorstores import Chroma
+from langchain_community.embeddings import SentenceTransformerEmbeddings
+import streamlit as st 
+
+st.set_page_config(
+    page_title="📚 Talk to Galvin's OS Textbook",
+    page_icon="👻",
+)
+load_dotenv()
+
+os.environ["HUGGINGFACEHUB_API_TOKEN"] = os.getenv("HUGGINGFACEHUB_API_TOKEN")
+
+prompt_template = """To provide the best response, consider the following context and question carefully:
+
+Context: {context}
+Question: {question}
+
+Provide an accurate and concise response based on the given context and question.
+"""
+
+prompt = PromptTemplate(template=prompt_template, input_variables=["question", "context"])
+
+embeddings = SentenceTransformerEmbeddings(model_name="llmware/industry-bert-insurance-v0.1")
+
+load_vector_store = Chroma(persist_directory="operatingsystem/embed", embedding_function=embeddings)
+
+lvs = load_vector_store.as_retriever(search_kwargs={"k":2})
+
+repo = "llmware/bling-sheared-llama-1.3b-0.1"
+
+hfllm = HuggingFaceHub(
+    repo_id=repo, model_kwargs={"temperature": 0.3, "max_length": 500}
+)
+
+kwargs_type = {"prompt": prompt}
+
+def quesans():
+    qa = RetrievalQA.from_chain_type(
+    llm=hfllm,
+    chain_type="stuff",
+    retriever=lvs,
+    return_source_documents=True,
+    chain_type_kwargs=kwargs_type,
+    verbose=True
+    )
+    return qa
+
+qa = quesans()
+
+def main():
+
+    st.title("📚 Talk to Galvin's OS Textbook ")
+
+    st.markdown(
+        """
+        <style>
+        body {
+            background-color: #FFC0CB; 
+        }
+        </style>
+        """,
+        unsafe_allow_html=True
+    )
+
+    text_query = st.text_area("Type your question here...", height=100)
+
+    generate_response_btn = st.button("Get Answer")
+
+    st.subheader("🎉 Answer 🎉")
+
+    if generate_response_btn and text_query:
+        with st.spinner("Generating response..."):
+            text_response = qa(text_query)
+            if text_response:
+                st.write(text_response)
+                st.success("Response generated!")
+            else:
+                st.error("Oops! I have no idea what you mean.")
+            st.balloons() 
+
+if __name__ == "__main__":
+    main()
+

Quest/OS_InsightX/requirements.txt

@@ -0,0 +1,10 @@
+torch
+sentence_transformers
+streamlit
+langchain 
+chromadb
+pypdf
+huggingface_hub
+python-dotenv
+
+langchain-community

Quest/OS_InsightX/vector.py

@@ -0,0 +1,18 @@
+import os
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings import SentenceTransformerEmbeddings
+from langchain.document_loaders import DirectoryLoader
+from langchain.document_loaders import PyPDFLoader
+from langchain.vectorstores import Chroma
+
+
+embeddings = SentenceTransformerEmbeddings(model_name="llmware/industry-bert-insurance-v0.1")
+
+loader = DirectoryLoader('data/', glob="**/*.pdf", show_progress=True, loader_cls=PyPDFLoader)
+documents = loader.load()
+text_splitter = RecursiveCharacterTextSplitter(chunk_size=700, chunk_overlap=70)
+texts = text_splitter.split_documents(documents)
+
+vector_store = Chroma.from_documents(texts, embeddings, collection_metadata={"hnsw:space": "cosine"}, persist_directory="stores/insurance_cosine")
+
+print("Vector DB Successfully Created!")