Spaces:
Running
on
T4
Running
on
T4
rerank model
Browse files
pages/Multimodal_Conversational_Search.py
CHANGED
@@ -34,9 +34,6 @@ st.set_page_config(
|
|
34 |
layout="wide",
|
35 |
page_icon="images/opensearch_mark_default.png"
|
36 |
)
|
37 |
-
if "trigger_search" not in st.session_state:
|
38 |
-
st.session_state.trigger_search = False
|
39 |
-
|
40 |
|
41 |
parent_dirname = "/".join((os.path.dirname(__file__)).split("/")[0:-1])
|
42 |
USER_ICON = "images/user.png"
|
@@ -45,11 +42,22 @@ REGENERATE_ICON = "images/regenerate.png"
|
|
45 |
s3_bucket_ = "pdf-repo-uploads"
|
46 |
#"pdf-repo-uploads"
|
47 |
|
48 |
-
|
49 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
|
51 |
|
52 |
# Check if the user ID is already stored in the session state
|
|
|
|
|
|
|
|
|
53 |
if 'user_id' in st.session_state:
|
54 |
user_id = st.session_state['user_id']
|
55 |
#print(f"User ID: {user_id}")
|
@@ -103,12 +111,6 @@ if "input_rag_searchType" not in st.session_state:
|
|
103 |
st.session_state.input_rag_searchType = ["Vector Search"]
|
104 |
|
105 |
|
106 |
-
|
107 |
-
region = 'us-east-1'
|
108 |
-
bedrock_runtime_client = boto3.client('bedrock-runtime',region_name=region)
|
109 |
-
output = []
|
110 |
-
service = 'es'
|
111 |
-
|
112 |
st.markdown("""
|
113 |
<style>
|
114 |
[data-testid=column]:nth-of-type(2) [data-testid=stVerticalBlock]{
|
@@ -154,7 +156,6 @@ if clear:
|
|
154 |
|
155 |
|
156 |
def handle_input():
|
157 |
-
st.session_state.trigger_search = True
|
158 |
print("Question: "+st.session_state.input_query)
|
159 |
print("-----------")
|
160 |
print("\n\n")
|
@@ -207,29 +208,33 @@ def render_answer(question,answer,index,res_img):
|
|
207 |
ans_ = answer['answer']
|
208 |
st.write(ans_)
|
209 |
|
210 |
-
polly_response = polly_client.synthesize_speech(VoiceId='Joanna',
|
211 |
-
|
212 |
-
|
213 |
-
|
214 |
|
215 |
-
audio_col1, audio_col2 = st.columns([50,50])
|
216 |
-
with audio_col1:
|
217 |
-
|
218 |
-
rdn_key_1 = ''.join([random.choice(string.ascii_letters)
|
219 |
-
|
220 |
# def show_maxsim():
|
221 |
# st.session_state.show_columns = True
|
222 |
# st.session_state.maxSimImages = colpali.img_highlight(st.session_state.top_img, st.session_state.query_token_vectors, st.session_state.query_tokens)
|
|
|
223 |
# with placeholder.container():
|
224 |
-
#
|
225 |
-
# handle_input()
|
226 |
-
# render_all()
|
227 |
-
# #render_all()
|
228 |
# if(st.session_state.input_is_colpali):
|
229 |
# st.button("Show similarity map",key=rdn_key_1,on_click = show_maxsim)
|
230 |
|
231 |
colu1,colu2,colu3 = st.columns([4,82,20])
|
232 |
with colu2:
|
|
|
|
|
|
|
|
|
|
|
|
|
233 |
with st.expander("Relevant Sources:"):
|
234 |
with st.container():
|
235 |
if(len(res_img)>0):
|
@@ -265,45 +270,42 @@ def render_answer(question,answer,index,res_img):
|
|
265 |
with cols[idx]:
|
266 |
|
267 |
st.image(parent_dirname+"/figures/"+st.session_state.input_index+"/"+img+".jpg")
|
268 |
-
#st.write(caption)
|
269 |
idx = idx+1
|
270 |
if(st.session_state.show_columns == True):
|
271 |
st.session_state.show_columns = False
|
272 |
-
#st.markdown("<div style='color:#e28743';padding:3px 7px 3px 7px;borderWidth: 0px;borderColor: red;borderStyle: solid;width: fit-content;height: fit-content;border-radius: 10px;'><b>Sources from the document:</b></div>", unsafe_allow_html = True)
|
273 |
if(len(answer["table"] )>0):
|
274 |
#with st.expander("Table:"):
|
275 |
-
df =
|
276 |
-
df.fillna(method='pad', inplace=True)
|
277 |
st.table(df)
|
278 |
#with st.expander("Raw sources:"):
|
279 |
st.write(answer["source"])
|
280 |
|
281 |
|
282 |
-
with col_3:
|
283 |
-
|
284 |
-
|
285 |
-
|
286 |
-
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
-
|
291 |
-
|
292 |
-
|
293 |
-
|
294 |
|
295 |
-
|
296 |
-
|
297 |
-
|
298 |
-
|
299 |
-
|
300 |
-
|
301 |
-
|
302 |
-
|
303 |
-
|
304 |
-
|
305 |
-
|
306 |
-
|
307 |
|
308 |
|
309 |
#Each answer will have context of the question asked in order to associate the provided feedback with the respective question
|
@@ -326,19 +328,21 @@ def render_all():
|
|
326 |
|
327 |
placeholder = st.empty()
|
328 |
with placeholder.container():
|
329 |
-
|
330 |
-
|
|
|
331 |
render_all()
|
332 |
-
|
333 |
|
334 |
st.markdown("")
|
335 |
col_2, col_3 = st.columns([75,20])
|
336 |
with col_2:
|
337 |
-
#st.markdown("")
|
338 |
input = st.text_input( "Ask here",label_visibility = "collapsed",key="input_query")
|
339 |
with col_3:
|
340 |
-
#
|
341 |
-
play = st.button("Go",
|
|
|
|
|
342 |
with st.sidebar:
|
343 |
st.page_link("app.py", label=":orange[Home]", icon="๐ ")
|
344 |
st.subheader(":blue[Sample Data]")
|
@@ -411,10 +415,5 @@ with st.sidebar:
|
|
411 |
with st.expander("Sample questions for Colpali retriever:"):
|
412 |
st.write("1. Proportion of female new hires 2021-2023? \n\n 2. First-half 2021 return on unlisted real estate investments? \n\n 3. Trend of the fund's expected absolute volatility between January 2014 and January 2016? \n\n 4. Fund return percentage in 2017? \n\n 5. Annualized gross return of the fund from 1997 to 2008?")
|
413 |
|
414 |
-
run = st.sidebar.button("๐ Run Search")
|
415 |
-
|
416 |
-
if run:
|
417 |
-
st.session_state.trigger_search = True
|
418 |
|
419 |
-
|
420 |
|
|
|
34 |
layout="wide",
|
35 |
page_icon="images/opensearch_mark_default.png"
|
36 |
)
|
|
|
|
|
|
|
37 |
|
38 |
parent_dirname = "/".join((os.path.dirname(__file__)).split("/")[0:-1])
|
39 |
USER_ICON = "images/user.png"
|
|
|
42 |
s3_bucket_ = "pdf-repo-uploads"
|
43 |
#"pdf-repo-uploads"
|
44 |
|
45 |
+
# @st.cache_resource
|
46 |
+
# def get_polly_client():
|
47 |
+
# return boto3.client('polly',
|
48 |
+
# aws_access_key_id=st.secrets['user_access_key'],
|
49 |
+
# aws_secret_access_key=st.secrets['user_secret_key'],
|
50 |
+
# region_name='us-east-1'
|
51 |
+
# )
|
52 |
+
|
53 |
+
# polly_client = get_polly_client()
|
54 |
|
55 |
|
56 |
# Check if the user ID is already stored in the session state
|
57 |
+
|
58 |
+
if "trigger_search" not in st.session_state:
|
59 |
+
st.session_state.trigger_search = False
|
60 |
+
|
61 |
if 'user_id' in st.session_state:
|
62 |
user_id = st.session_state['user_id']
|
63 |
#print(f"User ID: {user_id}")
|
|
|
111 |
st.session_state.input_rag_searchType = ["Vector Search"]
|
112 |
|
113 |
|
|
|
|
|
|
|
|
|
|
|
|
|
114 |
st.markdown("""
|
115 |
<style>
|
116 |
[data-testid=column]:nth-of-type(2) [data-testid=stVerticalBlock]{
|
|
|
156 |
|
157 |
|
158 |
def handle_input():
|
|
|
159 |
print("Question: "+st.session_state.input_query)
|
160 |
print("-----------")
|
161 |
print("\n\n")
|
|
|
208 |
ans_ = answer['answer']
|
209 |
st.write(ans_)
|
210 |
|
211 |
+
# polly_response = polly_client.synthesize_speech(VoiceId='Joanna',
|
212 |
+
# OutputFormat='ogg_vorbis',
|
213 |
+
# Text = ans_,
|
214 |
+
# Engine = 'neural')
|
215 |
|
216 |
+
# audio_col1, audio_col2 = st.columns([50,50])
|
217 |
+
# with audio_col1:
|
218 |
+
# st.audio(polly_response['AudioStream'].read(), format="audio/ogg")
|
219 |
+
# rdn_key_1 = ''.join([random.choice(string.ascii_letters)
|
220 |
+
# for _ in range(10)])
|
221 |
# def show_maxsim():
|
222 |
# st.session_state.show_columns = True
|
223 |
# st.session_state.maxSimImages = colpali.img_highlight(st.session_state.top_img, st.session_state.query_token_vectors, st.session_state.query_tokens)
|
224 |
+
# handle_input()
|
225 |
# with placeholder.container():
|
226 |
+
# render_all()
|
|
|
|
|
|
|
227 |
# if(st.session_state.input_is_colpali):
|
228 |
# st.button("Show similarity map",key=rdn_key_1,on_click = show_maxsim)
|
229 |
|
230 |
colu1,colu2,colu3 = st.columns([4,82,20])
|
231 |
with colu2:
|
232 |
+
@st.cache_data
|
233 |
+
def load_table_from_file(filepath):
|
234 |
+
df = pd.read_csv(filepath, skipinitialspace=True, on_bad_lines='skip', delimiter='`')
|
235 |
+
df.fillna(method='pad', inplace=True)
|
236 |
+
return df
|
237 |
+
|
238 |
with st.expander("Relevant Sources:"):
|
239 |
with st.container():
|
240 |
if(len(res_img)>0):
|
|
|
270 |
with cols[idx]:
|
271 |
|
272 |
st.image(parent_dirname+"/figures/"+st.session_state.input_index+"/"+img+".jpg")
|
|
|
273 |
idx = idx+1
|
274 |
if(st.session_state.show_columns == True):
|
275 |
st.session_state.show_columns = False
|
|
|
276 |
if(len(answer["table"] )>0):
|
277 |
#with st.expander("Table:"):
|
278 |
+
df = load_table_from_file(answer["table"][0]['name'])
|
|
|
279 |
st.table(df)
|
280 |
#with st.expander("Raw sources:"):
|
281 |
st.write(answer["source"])
|
282 |
|
283 |
|
284 |
+
# with col_3:
|
285 |
+
# if(index == len(st.session_state.questions_)):
|
286 |
+
|
287 |
+
# rdn_key = ''.join([random.choice(string.ascii_letters)
|
288 |
+
# for _ in range(10)])
|
289 |
+
# 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
|
290 |
+
# 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"])
|
291 |
+
# def on_button_click():
|
292 |
+
# if(currentValue!=oldValue or 1==1):
|
293 |
+
# st.session_state.input_query = st.session_state.questions_[-1]["question"]
|
294 |
+
# st.session_state.answers_.pop()
|
295 |
+
# st.session_state.questions_.pop()
|
296 |
|
297 |
+
# handle_input()
|
298 |
+
# with placeholder.container():
|
299 |
+
# render_all()
|
300 |
+
# if("currentValue" in st.session_state):
|
301 |
+
# del st.session_state["currentValue"]
|
302 |
+
|
303 |
+
# try:
|
304 |
+
# del regenerate
|
305 |
+
# except:
|
306 |
+
# pass
|
307 |
+
# placeholder__ = st.empty()
|
308 |
+
# placeholder__.button("๐",key=rdn_key,on_click=on_button_click)
|
309 |
|
310 |
|
311 |
#Each answer will have context of the question asked in order to associate the provided feedback with the respective question
|
|
|
328 |
|
329 |
placeholder = st.empty()
|
330 |
with placeholder.container():
|
331 |
+
if st.session_state.trigger_search:
|
332 |
+
with st.spinner("Running search..."):
|
333 |
+
handle_input()
|
334 |
render_all()
|
335 |
+
st.session_state.trigger_search = False # reset
|
336 |
|
337 |
st.markdown("")
|
338 |
col_2, col_3 = st.columns([75,20])
|
339 |
with col_2:
|
|
|
340 |
input = st.text_input( "Ask here",label_visibility = "collapsed",key="input_query")
|
341 |
with col_3:
|
342 |
+
#play = st.button("Go",on_click=handle_input,key = "play")
|
343 |
+
play = st.button("Go", key="play")
|
344 |
+
if play:
|
345 |
+
st.session_state.trigger_search = True
|
346 |
with st.sidebar:
|
347 |
st.page_link("app.py", label=":orange[Home]", icon="๐ ")
|
348 |
st.subheader(":blue[Sample Data]")
|
|
|
415 |
with st.expander("Sample questions for Colpali retriever:"):
|
416 |
st.write("1. Proportion of female new hires 2021-2023? \n\n 2. First-half 2021 return on unlisted real estate investments? \n\n 3. Trend of the fund's expected absolute volatility between January 2014 and January 2016? \n\n 4. Fund return percentage in 2017? \n\n 5. Annualized gross return of the fund from 1997 to 2008?")
|
417 |
|
|
|
|
|
|
|
|
|
418 |
|
|
|
419 |
|