Update app.py

app.py

@@ -15,20 +15,20 @@ from langchain_community.llms import HuggingFaceHub
 
 # define constants
 # Embedding models
-EMB_MODEL_bge_base = 'BAAI/bge-base-en-v1.5'
+#EMB_MODEL_bge_base = 'BAAI/bge-base-en-v1.5'
 EMB_MODEL_bge_large = 'BAAI/bge-large-en-v1.5'
-EMB_MODEL_gtr_t5_base = 'sentence-transformers/gtr-t5-base'
+#EMB_MODEL_gtr_t5_base = 'sentence-transformers/gtr-t5-base'
 EMB_MODEL_gtr_t5_large = 'sentence-transformers/gtr-t5-large'
-EMB_MODEL_e5_base = 'intfloat/e5-large-v2'
+#EMB_MODEL_e5_base = 'intfloat/e5-large-v2'
 # Chat app model
 MISTRAL_MODEL1 = 'mistralai/Mixtral-8x7B-Instruct-v0.1'
 HF_MODEL1 = 'HuggingFaceH4/zephyr-7b-beta'
 # define paths
-vector_path_bge_base = 'vectorDB/faiss_index_bge_base' 
+#vector_path_bge_base = 'vectorDB/faiss_index_bge_base' 
 vector_path_bge_large = 'vectorDB/faiss_index_bge_large'
-vector_path_gtr_t5_base = 'vectorDB/faiss_index_gtr_t5_base'
+#vector_path_gtr_t5_base = 'vectorDB/faiss_index_gtr_t5_base'
 vector_path_gtr_t5_large = 'vectorDB/faiss_index_gtr_t5_large'
-vector_path_e5_base = 'vectorDB/faiss_index_e5_base'
+#vector_path_e5_base = 'vectorDB/faiss_index_e5_base'
 
 hf_token = os.environ["HUGGINGFACEHUB_API_TOKEN"]
 
@@ -37,17 +37,17 @@ def respond(message, history):
     # Initialize your embedding model
     embedding_model_bge = HuggingFaceEmbeddings(model_name=EMB_MODEL_bge_large)
     embedding_model_gtr_t5 = HuggingFaceEmbeddings(model_name=EMB_MODEL_gtr_t5_large)
-    embedding_model_e5 = HuggingFaceEmbeddings(model_name=EMB_MODEL_e5_base)
+    #embedding_model_e5 = HuggingFaceEmbeddings(model_name=EMB_MODEL_e5_base)
     
     # Load FAISS from relative path
     vectordb_bge = FAISS.load_local(vector_path_bge_large, embedding_model_bge, allow_dangerous_deserialization=True)
     vectordb_gtr_t5 = FAISS.load_local(vector_path_gtr_t5_large, embedding_model_gtr_t5, allow_dangerous_deserialization=True)
-    vectordb_e5 = FAISS.load_local(vector_path_e5_base, embedding_model_e5, allow_dangerous_deserialization=True)
+    #vectordb_e5 = FAISS.load_local(vector_path_e5_base, embedding_model_e5, allow_dangerous_deserialization=True)
     
     # define retriever object
     retriever_bge = vectordb_bge.as_retriever(search_type="similarity", search_kwargs={"k": 5})
     retriever_gtr_t5 = vectordb_gtr_t5.as_retriever(search_type="similarity", search_kwargs={"k": 5})
-    retriever_e5 = vectordb_e5.as_retriever(search_type="similarity", search_kwargs={"k": 5})
+    #retriever_e5 = vectordb_e5.as_retriever(search_type="similarity", search_kwargs={"k": 5})
 
     # initialse chatbot llm
     llm = HuggingFaceHub(
@@ -59,12 +59,12 @@ def respond(message, history):
     # create a RAG pipeline
     qa_chain_bge = RetrievalQA.from_chain_type(llm=llm, retriever=retriever_bge)
     qa_chain_gtr_t5 = RetrievalQA.from_chain_type(llm=llm, retriever=retriever_gtr_t5)
-    qa_chain_e5 = RetrievalQA.from_chain_type(llm=llm, retriever=retriever_e5)
+    #qa_chain_e5 = RetrievalQA.from_chain_type(llm=llm, retriever=retriever_e5)
     
     #generate results
     responce_bge = qa_chain_bge.invoke(message)['result']
     responce_gtr_t5 = qa_chain_gtr_t5.invoke(message)['result']
-    responce_e5 = qa_chain_e5.invoke(message)['result']
+    #responce_e5 = qa_chain_e5.invoke(message)['result']
 
     # remove the top instructions
     #instruction_prefix = (
@@ -98,27 +98,37 @@ def respond(message, history):
     #    answer_e5 = match_e5.group(3).strip()
     #
     #formatted_responce = f'Question:{question_bge}\nHelpful Answer Type 1:\n{answer_bge}\nHelpful Answer Type 2:\n{answer_gtr_t5}\nHelpful Answer Type 3:\n{answer_e5}'
-    formatted_responce = f'\n*************  BAAI/bge-large-en-v1.5  ****************\n{responce_bge}\n**************  sentence-transformers/gtr-t5-large  ***************\n{responce_gtr_t5}\n************  intfloat/e5-large-v2  **************\n{responce_e5}'
+    #formatted_responce = f'\n*************  BAAI/bge-large-en-v1.5  ****************\n{responce_bge}\n**************  sentence-transformers/gtr-t5-large  ***************\n{responce_gtr_t5}\n************  intfloat/e5-large-v2  **************\n{responce_e5}'
+    formatted_responce = f'\n*************  BAAI/bge-large-en-v1.5  ****************\n{responce_bge}\n**************  sentence-transformers/gtr-t5-large  ***************\n{responce_gtr_t5}'
     yield formatted_responce
 
+with gr.Blocks() as demo:
+     gr.Markdown("# Intelligent Financial Document Q&A App")
+    
+    # About the App
+    with gr.Tab("About the App"):
+        gr.Markdown(about_lines)
 
-demo = gr.ChatInterface(
-    respond,
-    type="messages", 
-    autofocus=False #,
-    #additional_inputs=[
-    #    gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-    #    gr.Slider(minimum=128, maximum=1024, value=512, step=128, label="Max new tokens"),
-    #    gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature"),
-    #    gr.Slider(
-    #        minimum=0.1,
-    #        maximum=1.0,
-    #        value=0.95,
-    #        step=0.05,
-    #        label="Top-p (nucleus sampling)",
-    #    ),
-    #],
-)
+    # Document Chatbot
+        with gr.Tab("Market Prediction"):
+            #demo = gr.ChatInterface(
+            gr.ChatInterface(
+                respond,
+                type="messages", 
+                autofocus=False #,
+                #additional_inputs=[
+                #    gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+                #    gr.Slider(minimum=128, maximum=1024, value=512, step=128, label="Max new tokens"),
+                #    gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature"),
+                #    gr.Slider(
+                #        minimum=0.1,
+                #        maximum=1.0,
+                #        value=0.95,
+                #        step=0.05,
+                #        label="Top-p (nucleus sampling)",
+                #    ),
+                #],
+            )
 
 
 if __name__ == "__main__":