Gabriela Nicole Gonzalez Saez commited on
Commit
0ace0a9
1 Parent(s): 1115254

fix empty start

Browse files
Files changed (1) hide show
  1. app.py +2 -2
app.py CHANGED
@@ -129,7 +129,7 @@ def split_token_from_sequences(sequences, model) -> dict :
129
  cont = False
130
  step_words = []
131
  for beam in range(n_sentences):
132
- app_text = ''
133
  if step_i < len(gen_sequences_texts[beam]):
134
  app_text = gen_sequences_texts[beam][step_i]
135
  cont = True
@@ -138,7 +138,7 @@ def split_token_from_sequences(sequences, model) -> dict :
138
  print(words_by_step)
139
 
140
  for i_bs, step_w in enumerate(step_words):
141
- if step_w != '':
142
  #new id if the same word is not in another beam (?) [beam[i] was a token id]
143
  #parent id = previous word and previous step.
144
 
 
129
  cont = False
130
  step_words = []
131
  for beam in range(n_sentences):
132
+ app_text = '<empty_word>'
133
  if step_i < len(gen_sequences_texts[beam]):
134
  app_text = gen_sequences_texts[beam][step_i]
135
  cont = True
 
138
  print(words_by_step)
139
 
140
  for i_bs, step_w in enumerate(step_words):
141
+ if not step_w in ['<empty_word>', '<pad>']:
142
  #new id if the same word is not in another beam (?) [beam[i] was a token id]
143
  #parent id = previous word and previous step.
144