rerank model

RAG/rag_DocumentSearcher.py

@@ -12,7 +12,7 @@ headers = {"Content-Type": "application/json"}
 host = "https://search-opensearchservi-shjckef2t7wo-iyv6rajdgxg6jas25aupuxev6i.us-west-2.es.amazonaws.com/"
 
 parent_dirname = "/".join((os.path.dirname(__file__)).split("/")[0:-1])
-
+@st.cache_data
 def query_(awsauth,inputs, session_id,search_types):
 
     print("using index: "+st.session_state.input_index)

pages/Multimodal_Conversational_Search.py

@@ -39,6 +39,7 @@ AI_ICON = "images/opensearch-twitter-card.png"
 REGENERATE_ICON = "images/regenerate.png"
 s3_bucket_ = "pdf-repo-uploads"
             #"pdf-repo-uploads"
+@st.cache_data
 polly_client = boto3.client('polly',aws_access_key_id=st.secrets['user_access_key'],
                 aws_secret_access_key=st.secrets['user_secret_key'], region_name = 'us-east-1')
 
@@ -99,6 +100,7 @@ if "input_rag_searchType" not in st.session_state:
 
         
 region = 'us-east-1'
+@st.cache_data
 bedrock_runtime_client = boto3.client('bedrock-runtime',region_name=region)
 output = []
 service = 'es'
@@ -345,8 +347,6 @@ def render_answer(question,answer,index,res_img):
 
             rdn_key = ''.join([random.choice(string.ascii_letters)
                               for _ in range(10)])
-            # rdn_key_1 = ''.join([random.choice(string.ascii_letters)
-            #                   for _ in range(10)])
             currentValue = ''.join(st.session_state.input_rag_searchType)+str(st.session_state.input_is_rerank)+str(st.session_state.input_table_with_sql)+st.session_state.input_index
             oldValue = ''.join(st.session_state.inputs_["rag_searchType"])+str(st.session_state.inputs_["is_rerank"])+str(st.session_state.inputs_["table_with_sql"])+str(st.session_state.inputs_["index"])
             def on_button_click():
@@ -358,12 +358,6 @@ def render_answer(question,answer,index,res_img):
                     handle_input()
                     with placeholder.container():
                         render_all()
-            # def show_maxsim():
-            #     st.session_state.show_columns = True
-            #     st.session_state.maxSimImages = colpali.img_highlight(st.session_state.top_img, st.session_state.query_token_vectors, st.session_state.query_tokens)
-            #     handle_input()
-            #     with placeholder.container():
-            #         render_all()
             if("currentValue"  in st.session_state):
                 del st.session_state["currentValue"]
 

utilities/invoke_models.py

@@ -11,6 +11,7 @@ import streamlit as st
 #import torch
 
 region = 'us-east-1'
+@st.cache_data
 bedrock_runtime_client = boto3.client(
     'bedrock-runtime',
     aws_access_key_id=st.secrets['user_access_key'],
@@ -29,7 +30,7 @@ bedrock_runtime_client = boto3.client(
 #     max_length = 16
 #     num_beams = 4
 #     gen_kwargs = {"max_length": max_length, "num_beams": num_beams}
-
+@st.cache_data
 def invoke_model(input):
     response = bedrock_runtime_client.invoke_model(
         body=json.dumps({
@@ -42,7 +43,7 @@ def invoke_model(input):
     
     response_body = json.loads(response.get("body").read())
     return response_body.get("embedding")
-
+@st.cache_data
 def invoke_model_mm(text,img):
     body_ = {
             "inputText": text,
@@ -63,7 +64,7 @@ def invoke_model_mm(text,img):
     response_body = json.loads(response.get("body").read())
     #print(response_body)
     return response_body.get("embedding")
-
+@st.cache_data
 def invoke_llm_model(input,is_stream):
     if(is_stream == False):
         response = bedrock_runtime_client.invoke_model( 
@@ -144,7 +145,7 @@ def invoke_llm_model(input,is_stream):
     #     stream = response.get('body')
         
     #     return stream
-        
+@st.cache_data        
 def read_from_table(file,question):
     print("started table analysis:")
     print("-----------------------")
@@ -159,7 +160,7 @@ def read_from_table(file,question):
     "top_p":0.7,
     "stop_sequences":["\\n\\nHuman:"] 
     }
-    
+    @st.cache_data
     model = BedrockChat(
     client=bedrock_runtime_client,
     model_id='anthropic.claude-3-sonnet-20240229-v1:0',
@@ -180,7 +181,7 @@ def read_from_table(file,question):
              )
     agent_res = agent.invoke(question)['output']
     return agent_res
-    
+@st.cache_data    
 def generate_image_captions_llm(base64_string,question):
     
     # ant_client = Anthropic()