Update app.py

app.py

@@ -5,7 +5,6 @@ from youtube_transcript_api import YouTubeTranscriptApi
 import os
 import ast 
 import pandas as pd
-import before_run 
 from segmentation import SemanticTextSegmentation
 import re
 from symspellpy import SymSpell
@@ -13,9 +12,12 @@ import pkg_resources
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 from torch import cuda 
 from transformers import pipeline
+import nltk
+nltk.download('stopwords')
 from PIL import Image
 from PIL import ImageDraw
 from PIL import ImageFont
+import time
 
 
 if not os.path.exists('./transcripts'):
@@ -46,7 +48,6 @@ def clean_text(link,start,end):
     #print(starts)
     #ts_ = float(ts.strip("s&end"))
     #es_ = float(es.strip(es[-1]))
-    st.write('this is the value of es: ',es)
     if not(es) :
         e_val = starts[-1]
     else:
@@ -146,13 +147,12 @@ def clean_text(link,start,end):
     # remove Nan
     df.dropna(inplace=True)
     sts = SemanticTextSegmentation(df)
-    texts,i_r = sts.get_segments()
-    return texts,i_r
+    texts = sts.get_segments()
+    return texts
   
   sf = pd.DataFrame(columns=['Segmented_Text','video_id'])
 
-  text,i_array = segment(transcript.at[0,'text'])
-  st.write(i_array)
+  text = segment(transcript.at[0,'text'])
   for i in range(len(text)):
     sf.loc[i, 'Segmented_Text'] = text[i]
     sf.loc[i, 'video_id'] = transcript.at[0,'video_id']
@@ -201,8 +201,11 @@ def clean_text(link,start,end):
 def t5_summarizer(link,start, end):
     input_text = clean_text(link,start,end)
     lst_outputs = []
-    tokenizer1 = AutoTokenizer.from_pretrained("CareerNinja/t5-large_3e-4")
-    model1 = AutoModelForSeq2SeqLM.from_pretrained("CareerNinja/t5-large_3e-4")
+    tokenizer1 = AutoTokenizer.from_pretrained("CareerNinja/t5_large_3e-4_on_v2_dataset")
+    st.write('Loading the model!')
+    start_time = time.time()
+    model1 = AutoModelForSeq2SeqLM.from_pretrained("CareerNinja/t5_large_3e-4_on_v2_dataset")
+    st.write('Model loading compelete, time taken: ',time.time()-start_time)
     summarizer1 = pipeline("summarization", model=model1, tokenizer=tokenizer1)
     print(f""" Entered summarizer ! """)
     st.write('Below is the summary of the given URL: ')