Update app.py

app.py

@@ -26,7 +26,7 @@ import transformers
 from pydub import AudioSegment
 from streamlit_extras.streaming_write import write
 import time
-from langchain.llms import HuggingFaceHub
+
 import transformers
 from transformers import MBartForConditionalGeneration, MBart50TokenizerFast
 translation_model = MBartForConditionalGeneration.from_pretrained("facebook/mbart-large-50-many-to-many-mmt")
@@ -123,10 +123,8 @@ def load_model(_docs):
                              top_p=0.95,
                              repetition_penalty=1.15,
                              streamer=streamer,)
-    #llm = HuggingFacePipeline(pipeline=text_pipeline, model_kwargs={"temperature": 0.1})
-    
-    llm = HuggingFaceHub(repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1", model_kwargs={"temperature": 0.1, "max_new_tokens": 1024, "top_k": 3, "load_in_8bit": True})
-    
+    llm = HuggingFacePipeline(pipeline=text_pipeline, model_kwargs={"temperature": 0.1})
+
     # SYSTEM_PROMPT = ("Use the following pieces of context to answer the question at the end. "
     #                  "If you don't know the answer, just say that you don't know, "
     #                  "don't try to make up an answer.")
@@ -139,7 +137,7 @@ def load_model(_docs):
     qa_chain = RetrievalQA.from_chain_type(
         llm=llm,
         chain_type="stuff",
-        retriever=db.as_retriever(search_kwargs={"k": 10}),
+        retriever=db.as_retriever(search_kwargs={"k": 3}),
         return_source_documents=True,
         chain_type_kwargs={"prompt": prompt,
                            "verbose": False})
@@ -178,16 +176,14 @@ qa_chain = load_model(docs)
 if prompt := st.chat_input("How can I help you today?"):
     st.session_state.messages.append({"role": "user", "content": prompt})
     with st.chat_message("user"):
-        #english_prompt = hindi_to_english(prompt)
-        #st.markdown(english_prompt)
-        st.markdown(prompt)
+        english_prompt = hindi_to_english(prompt)
+        st.markdown(english_prompt)
     with st.chat_message("assistant"):
         with st.spinner(text="Looking for relevant answer"):
             message_placeholder = st.empty()
             full_response = ""
             message_history = "\n".join(list(get_message_history())[-3:])
             result = qa_chain(prompt)
-            #result = qa_chain(english_prompt)
             output = [result['result']]
 
     def generate_pdf():
@@ -240,10 +236,10 @@ if prompt := st.chat_input("How can I help you today?"):
     #     #yield word + " "
     #     time.sleep(0.1)
     
-    #for item in output:
-     #   full_response += english_to_hindi(item)[0]
-      #  message_placeholder.markdown(full_response + "▌")
-      #  message_placeholder.markdown(full_response)
+    for item in output:
+        full_response += english_to_hindi(item)[0]
+        message_placeholder.markdown(full_response + "▌")
+        message_placeholder.markdown(full_response)
         # message_placeholder.markdown(result['source_documents'])
     
     #stream_example()
@@ -254,9 +250,7 @@ if prompt := st.chat_input("How can I help you today?"):
     #     message_placeholder.markdown(write(stream_example))
 
     #write(stream_example)
-    #message_placeholder.markdown(english_to_hindi(result['result'])[0])
-    message_placeholder.markdown(result['result'])
-
+    message_placeholder.markdown(english_to_hindi(result['result'])[0])
 
     # sound_file = BytesIO()
     # tts = gTTS(result['result'], lang='en')