theekshana
commited on
Commit
•
400b4d5
1
Parent(s):
06b581b
rouge score
Browse files- app.py +105 -5
- requirements.txt +0 -0
- rouge_evaluate.py +8 -0
app.py
CHANGED
@@ -6,6 +6,7 @@ import nltk
|
|
6 |
import validators
|
7 |
import streamlit as st
|
8 |
from summarizer import summarizer_init, summarizer_summarize
|
|
|
9 |
from config import MODELS
|
10 |
from warnings import filterwarnings
|
11 |
|
@@ -28,6 +29,11 @@ def initialize_app():
|
|
28 |
"model_name": "Boardpac summarizer v1",
|
29 |
"summarizer_type": "Map Reduce",
|
30 |
"is_parameters_changed":False,
|
|
|
|
|
|
|
|
|
|
|
31 |
# "user_question":'',
|
32 |
'openai_api_key':'',
|
33 |
}
|
@@ -155,6 +161,7 @@ def load_app():
|
|
155 |
# inp_text = st.text_input(
|
156 |
# "Enter text or a url here"
|
157 |
# )
|
|
|
158 |
inp_text = st.text_area(
|
159 |
"Enter text here"
|
160 |
)
|
@@ -177,7 +184,7 @@ def load_app():
|
|
177 |
cleaned_txt = clean_text(cleaned_txt)
|
178 |
else:
|
179 |
logger.info("Text Input Type: INPUT TEXT")
|
180 |
-
cleaned_txt = clean_text(inp_text)
|
181 |
|
182 |
# view summarized text (expander)
|
183 |
with st.expander("View input text"):
|
@@ -186,7 +193,14 @@ def load_app():
|
|
186 |
else:
|
187 |
st.write(cleaned_txt)
|
188 |
|
|
|
|
|
|
|
|
|
|
|
|
|
189 |
submitted = st.button("Summarize")
|
|
|
190 |
|
191 |
if submitted:
|
192 |
if is_url:
|
@@ -194,11 +208,61 @@ def load_app():
|
|
194 |
else:
|
195 |
text_to_summarize = cleaned_txt
|
196 |
|
197 |
-
submit_text_to_summarize(text_to_summarize)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
198 |
|
199 |
def submit_text_to_summarize(text_to_summarize):
|
200 |
summarized_text, time = get_summary(text_to_summarize)
|
|
|
201 |
display_output(summarized_text,time)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
202 |
|
203 |
|
204 |
def get_summary(text_to_summarize):
|
@@ -221,21 +285,57 @@ def get_summary(text_to_summarize):
|
|
221 |
summarized_text, time = summarizer_summarize(model_type,tokenizer, base_summarizer, text_to_summarize ,summarizer_type = "map_reduce")
|
222 |
return summarized_text, time
|
223 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
224 |
|
225 |
def display_output(summarized_text,time):
|
226 |
logger.info(f"SUMMARY: {summarized_text}")
|
227 |
logger.info(f"Summary took {time}s")
|
228 |
st.subheader("Summarized text")
|
229 |
st.info(f"{summarized_text}")
|
230 |
-
st.
|
231 |
-
|
232 |
|
233 |
def main():
|
234 |
|
235 |
initialize_app()
|
236 |
side_bar()
|
237 |
load_app()
|
238 |
-
# chat_body()
|
239 |
|
240 |
|
241 |
if __name__ == "__main__":
|
|
|
6 |
import validators
|
7 |
import streamlit as st
|
8 |
from summarizer import summarizer_init, summarizer_summarize
|
9 |
+
from rouge_evaluate import get_rouge_scores
|
10 |
from config import MODELS
|
11 |
from warnings import filterwarnings
|
12 |
|
|
|
29 |
"model_name": "Boardpac summarizer v1",
|
30 |
"summarizer_type": "Map Reduce",
|
31 |
"is_parameters_changed":False,
|
32 |
+
"is_evaluate_pressed":False,
|
33 |
+
"is_evaluate_pressed":False,
|
34 |
+
"reference_summary":'',
|
35 |
+
"generated_summary":'',
|
36 |
+
"summary_time":'',
|
37 |
# "user_question":'',
|
38 |
'openai_api_key':'',
|
39 |
}
|
|
|
161 |
# inp_text = st.text_input(
|
162 |
# "Enter text or a url here"
|
163 |
# )
|
164 |
+
# with st.form("Text Summarizer"):
|
165 |
inp_text = st.text_area(
|
166 |
"Enter text here"
|
167 |
)
|
|
|
184 |
cleaned_txt = clean_text(cleaned_txt)
|
185 |
else:
|
186 |
logger.info("Text Input Type: INPUT TEXT")
|
187 |
+
cleaned_txt = inp_text # clean_text(inp_text)
|
188 |
|
189 |
# view summarized text (expander)
|
190 |
with st.expander("View input text"):
|
|
|
193 |
else:
|
194 |
st.write(cleaned_txt)
|
195 |
|
196 |
+
st.subheader('Optional - Evaluate summary against a reference')
|
197 |
+
# with st.form('Evaluate summary against a reference'):
|
198 |
+
reference_summary = st.text_area(
|
199 |
+
"Enter reference summary here"
|
200 |
+
)
|
201 |
+
|
202 |
submitted = st.button("Summarize")
|
203 |
+
# submitted = st.form_submit_button("Summarize")
|
204 |
|
205 |
if submitted:
|
206 |
if is_url:
|
|
|
208 |
else:
|
209 |
text_to_summarize = cleaned_txt
|
210 |
|
211 |
+
summarized_text=submit_text_to_summarize(text_to_summarize)
|
212 |
+
# summarized_text=submit_text_to_summarize2(text_to_summarize, reference_summary)
|
213 |
+
# "reference_summary":'',
|
214 |
+
# st.session_state.generated_summary=summarized_text
|
215 |
+
# evaluate_block(summarized_text)
|
216 |
+
# st.subheader('Evaluate summary against a reference')
|
217 |
+
# with st.form('Evaluate summary against a reference'):
|
218 |
+
# reference_summary = st.text_area(
|
219 |
+
# "Enter reference summary here"
|
220 |
+
# )
|
221 |
+
# Evaluate = st.button(
|
222 |
+
# "Evaluate",
|
223 |
+
# # on_click=update_evaluate_button_change,
|
224 |
+
# on_click=testtttt,
|
225 |
+
# # args=[summarized_text, reference_summary]
|
226 |
+
# )
|
227 |
+
|
228 |
+
# if Evaluate :
|
229 |
+
if reference_summary.strip() != '':
|
230 |
+
summarized_text=st.session_state.generated_summary
|
231 |
+
rouge_result = get_rouge_scores(summarized_text, clean_text(reference_summary))
|
232 |
+
# st.text(f"evaluate scores-----: {scores}")
|
233 |
+
# st.info("evaluate scores-----:")
|
234 |
+
# st.info("evaluated scores-----:")
|
235 |
+
# with st.expander(f"evaluated scores: rouge1: {scores['rouge1']:.2f}%"):
|
236 |
+
# st.table(scores)
|
237 |
+
st.text("Evaluated scores:")
|
238 |
+
col1, col2, col3 = st.columns(3)
|
239 |
+
col1.metric('Rogue1', "{:.2f}".format(rouge_result['rouge1']))
|
240 |
+
col2.metric('rouge2', "{:.2f}".format(rouge_result['rouge2']))
|
241 |
+
col3.metric('rougeL', "{:.2f}".format(rouge_result['rougeL']))
|
242 |
+
# col4.metric('rougeLsum', "{:.2f}".format(rouge_result['rougeLsum']))
|
243 |
+
|
244 |
+
|
245 |
|
246 |
def submit_text_to_summarize(text_to_summarize):
|
247 |
summarized_text, time = get_summary(text_to_summarize)
|
248 |
+
st.session_state.generated_summary=summarized_text
|
249 |
display_output(summarized_text,time)
|
250 |
+
# evaluate_block(summarized_text)
|
251 |
+
# return summarized_text
|
252 |
+
|
253 |
+
def submit_text_to_summarize2(text_to_summarize, reference_summary):
|
254 |
+
summarized_text, time = get_summary(text_to_summarize)
|
255 |
+
# display_output(summarized_text,time)
|
256 |
+
logger.info(f"SUMMARY: {summarized_text}")
|
257 |
+
logger.info(f"Summary took {time}s")
|
258 |
+
st.subheader("Summarized text")
|
259 |
+
st.info(f"{summarized_text}")
|
260 |
+
# st.info(f"Time: {time}s")
|
261 |
+
st.text(f"Time taken: {time}s")
|
262 |
+
|
263 |
+
# scores = get_rouge_scores(summarized_text, reference_summary)
|
264 |
+
# st.markdown(f"evaluate scores: {scores}")
|
265 |
+
|
266 |
|
267 |
|
268 |
def get_summary(text_to_summarize):
|
|
|
285 |
summarized_text, time = summarizer_summarize(model_type,tokenizer, base_summarizer, text_to_summarize ,summarizer_type = "map_reduce")
|
286 |
return summarized_text, time
|
287 |
|
288 |
+
def evaluate_block(summarized_text):
|
289 |
+
st.subheader('Evaluate summary against a reference')
|
290 |
+
# with st.form('Evaluate summary against a reference'):
|
291 |
+
reference_summary = st.text_area(
|
292 |
+
"Enter reference summary here"
|
293 |
+
)
|
294 |
+
Evaluate = st.button(
|
295 |
+
"Evaluate",
|
296 |
+
on_click=testtttt,
|
297 |
+
args=[summarized_text,reference_summary]
|
298 |
+
)
|
299 |
+
|
300 |
+
# Evaluate = st.form_submit_button(
|
301 |
+
# "Evaluate key",
|
302 |
+
# # on_click=update_parameters_change
|
303 |
+
# on_click=testtttt,
|
304 |
+
# args=[summarized_text,reference_summary]
|
305 |
+
# )
|
306 |
+
# if Evaluate or st.session_state.is_evaluate_pressed:
|
307 |
+
# if reference_summary:
|
308 |
+
# # if
|
309 |
+
# scores = get_rouge_scores(summarized_text, reference_summary)
|
310 |
+
# st.text(f"evaluate scores: {scores}")
|
311 |
+
# st.session_state.is_evaluate_pressed = False
|
312 |
+
|
313 |
+
def testtttt(summarized_text, reference_summary):
|
314 |
+
print(summarized_text, reference_summary)
|
315 |
+
scores = get_rouge_scores(summarized_text, reference_summary)
|
316 |
+
st.text(f"evaluate scores-----: {scores}")
|
317 |
+
|
318 |
+
|
319 |
+
def update_evaluate_button_change():
|
320 |
+
st.session_state.is_evaluate_pressed = True
|
321 |
+
|
322 |
+
# def evaluate(summarized_text, reference_summary):
|
323 |
+
# return get_rouge_scores(summarized_text, reference_summary)
|
324 |
+
|
325 |
|
326 |
def display_output(summarized_text,time):
|
327 |
logger.info(f"SUMMARY: {summarized_text}")
|
328 |
logger.info(f"Summary took {time}s")
|
329 |
st.subheader("Summarized text")
|
330 |
st.info(f"{summarized_text}")
|
331 |
+
# st.info(f"Time: {time}s")
|
332 |
+
st.text(f"Time taken: {time}s")
|
333 |
|
334 |
def main():
|
335 |
|
336 |
initialize_app()
|
337 |
side_bar()
|
338 |
load_app()
|
|
|
339 |
|
340 |
|
341 |
if __name__ == "__main__":
|
requirements.txt
CHANGED
Binary files a/requirements.txt and b/requirements.txt differ
|
|
rouge_evaluate.py
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import evaluate
|
2 |
+
rouge_score = evaluate.load("rouge")
|
3 |
+
|
4 |
+
def get_rouge_scores(genererated_summary:str,reference_summary:str)->dict:
|
5 |
+
|
6 |
+
rouge_result = rouge_score.compute(predictions=[genererated_summary], references=[reference_summary])
|
7 |
+
|
8 |
+
return rouge_result
|