Spaces:
Running
Running
antoinelouis
commited on
Commit
•
70f0431
1
Parent(s):
0209ea7
Update app.py
Browse files
app.py
CHANGED
@@ -45,7 +45,7 @@ def estimate_pruned_vocabulary(tokenizer: PreTrainedTokenizerFast, language: str
|
|
45 |
counter = Counter(tokenizer.all_special_tokens)
|
46 |
for i, text in enumerate(df.text):
|
47 |
counter.update(tok for tok in tokenizer.tokenize(text))
|
48 |
-
my_bar.progress(i/len(df)
|
49 |
return set(counter)
|
50 |
else:
|
51 |
raise FileNotFoundError
|
|
|
45 |
counter = Counter(tokenizer.all_special_tokens)
|
46 |
for i, text in enumerate(df.text):
|
47 |
counter.update(tok for tok in tokenizer.tokenize(text))
|
48 |
+
my_bar.progress(i/len(df), text=f"{i/len(df)*100:.0f}%")
|
49 |
return set(counter)
|
50 |
else:
|
51 |
raise FileNotFoundError
|