added duckduckgo search

.gitignore

@@ -0,0 +1,3 @@
+venv
+.env
+__pycache__

app.py

@@ -1,6 +1,15 @@
 import streamlit as st
 from utils import st_load_retriever, st_load_llm, StreamHandler
 from langchain.chains import RetrievalQAWithSourcesChain
+from langchain.callbacks import StreamlitCallbackHandler
+from langchain.agents import AgentType, initialize_agent, load_tools
+
+
+RETRIEVAL_METHOD_MAP = {
+    "Vector Database": "vectordb",
+    "Google Search": "google search",
+    "DuckDuckGo Search": "duckduckgo search",
+}
 
 st.title("AIxplorer - A Smarter Google Scholar 🌐📚")
 st.write(
@@ -13,28 +22,42 @@ st.subheader("Settings")
 col1, col2, col3 = st.columns(3)
 
 with col1:
-    use_google = st.checkbox(
-        "Use Google Search",
-        value=True,
-        help="Use Google Search to retrieve papers. If unchecked, will use the vector database.",
+    retrieval_method = st.selectbox(
+        "Retrieval Mode",
+        RETRIEVAL_METHOD_MAP.keys(),
+        index=0,
+        help="The retrieval method used to retrieve supporting documents.",
     )
+
 st.divider()
 
 llm = st_load_llm()
-retriever = st_load_retriever(llm, "vectordb" if not use_google else "google search")
 
-qa_chain = RetrievalQAWithSourcesChain.from_chain_type(llm, retriever=retriever)
-
-user_input = st.text_area(
-    "Enter your query here",
-    help="Query should be on computer science as the RAG system is tuned to that domain.",
-)
-
-
-if st.button("Generate"):
-    st.divider()
-    st.subheader("Answer:")
-    with st.spinner("Generating..."):
-        container = st.empty()
-        stream_handler = StreamHandler(container)
-        response = qa_chain({"question": user_input}, callbacks=[stream_handler])
+# first path
+if retrieval_method in ("Vector Database", "Google Search"):
+    retriever = st_load_retriever(llm, RETRIEVAL_METHOD_MAP[retrieval_method])
+    qa_chain = RetrievalQAWithSourcesChain.from_chain_type(llm, retriever=retriever)
+    user_input = st.text_area(
+        "Enter your query here",
+        help="Query should be on computer science as the RAG system is tuned to that domain.",
+    )
+    if st.button("Generate"):
+        st.divider()
+        st.subheader("Answer:")
+        with st.spinner("Generating..."):
+            container = st.empty()
+            stream_handler = StreamHandler(container)
+            response = qa_chain({"question": user_input}, callbacks=[stream_handler])
+
+# second path
+else:
+    tools = load_tools(["ddg-search"])
+    agent = initialize_agent(
+        tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True
+    )
+    if prompt := st.chat_input():
+        st.chat_message("user").write(prompt)
+        with st.chat_message("assistant"):
+            st_callback = StreamlitCallbackHandler(st.container())
+            response = agent.run(prompt, callbacks=[st_callback])
+            st.write(response)

requirements.txt

@@ -11,14 +11,17 @@ beautifulsoup4==4.12.2
 blinker==1.7.0
 cachetools==5.3.2
 certifi==2023.11.17
+cffi==1.16.0
 charset-normalizer==3.3.2
 chroma-hnswlib==0.7.3
 chromadb==0.4.20
 click==8.1.7
 coloredlogs==15.0.1
+curl-cffi==0.5.10
 dataclasses-json==0.6.3
 Deprecated==1.2.14
 diskcache==5.6.3
+duckduckgo-search==4.1.0
 fastapi==0.105.0
 filelock==3.13.1
 flatbuffers==23.5.26
@@ -54,6 +57,7 @@ langchain-community==0.0.5
 langchain-core==0.1.2
 langsmith==0.0.72
 llama_cpp_python==0.2.24
+lxml==4.9.4
 markdown-it-py==3.0.0
 MarkupSafe==2.1.3
 marshmallow==3.20.1
@@ -100,6 +104,7 @@ pulsar-client==3.3.0
 pyarrow==14.0.2
 pyasn1==0.5.1
 pyasn1-modules==0.3.0
+pycparser==2.21
 pydantic==2.5.2
 pydantic_core==2.14.5
 pydeck==0.8.1b0

utils.py

@@ -21,7 +21,7 @@ class StreamHandler(BaseCallbackHandler):
 
 @st.cache_resource
 def st_load_retriever(_llm, mode):
-    model_kwargs = {"device": "cuda"}
+    model_kwargs = {"device": config.device}
     embeddings_model = HuggingFaceEmbeddings(
         model_name=config.embeddings_model,
         model_kwargs=model_kwargs,