How to fix tqdm progress_apply for pandas in Jupyter?

Question

Don't really understand is it a mistake or just my local problem, still have some issues with using tqdm progress bars with progress_apply in Jupyter.

First try:

from tqdm import tqdm
tqdm_notebook.pandas(desc="Example Desc")
keywords_df['keyword'] = keywords_df['keywird'].progress_apply(lambda x: x.replace('*',''))

Output (without any bars):

AttributeError: 'function' object has no attribute 'pandas'

Second try:

from tqdm import tqdm
tqdm_notebook().pandas(desc="Example Desc")
keywords_df['keyword'] = keywords_df['keywird'].progress_apply(lambda x: x.replace('*',''))

Output: Two bars (need one). First bar is empty (0it [00:00, ?it/s]), second is OK.

Any ideas how to change progress_apply description and display bar without empty initialization bar? :)

P.S. Documentation (https://github.com/tqdm/tqdm) says I can just use tqdm_notebook, but it's not working for me :)

# Register `pandas.progress_apply` and `pandas.Series.map_apply` with `tqdm`
# (can use `tqdm_gui`, `tqdm_notebook`, optional kwargs, etc.)
tqdm.pandas(desc="my bar!")

Answer 1

Answer from tqdm developer:

notebook support is still in a (late) beta stage. The API might change slightly when we release tqdm v5 but for now you probably need

from tqdm._tqdm_notebook import tqdm_notebook

tqdm_notebook.pandas(...