airbert-vln / bnb-dataset Goto Github PK

As you mentioned in the README.md file, Airbnb server blocks my request after certain amount of consecutive requests. You mentioned utilizing pool of ip adresses. Could you also address how to get a pool of ip addresses, and how to use them?
I know some tricks to circumvent the TooManyRequest error, but I wonder what your method is.

how can I get this file?

As far as I know, python search_listings.py --locations data/cities.txt --output data/listings creates multiple listings.txt files at multiple different newly created folders.

So, the code in python download_listings.py --listings data/listings.txt --output data/merlin does not make sense and raises error.
(there is no data/listings.txt after executing python search_listings.py --locations data/cities.txt --output data/listings.)
I'm using os.walk to adjust download_listings.py code in order to get the json files

Is it what you intended to do that python search_listings.py --locations data/cities.txt --output data/listings creates multiple listings.txt files?

⛓️ 5.1. Concatenation

In this paper, concatenation means Concatenating Images and Texts in a BnB Listing, it should exclude captionless images.

However, in "python preprocess_dataset.py --csv data/bnb-train.tsv --name bnb_train", parameter [--captionless] defaults to "True".

Should parameter [--captionless] be set "False" in this part?

👥 5.2. Image merging

"python merge_photos.py --source bnb_train.py --output merge+bnb_train.py --detection-dir data/places365 " has some errors.

Should [source] be set to "bnb_train.json" without captionless images or "2capt+bnb_train.json" with captionless images?

👣 5.4. Instruction rephrasing

In this part, I get "bnb-train.np.tsv" instead of "np+bnb_train.json" which is used in airbert training.

Could you explain these? Thanks.

Hi,

Right after cloning this repo, there are airbnb-test-indoor-filtered.tsv and airbnb-train-indoor-filtered.tsv in data folder.
But after following README.md, bnb-dataset-indoor.tsv, bnb-dataset-raw.tsv, bnb-dataset.tsv and noun_phrases.part-0-airbnb{train|test}.tsv(I named these files myself. Originally named as noun_phrases.part-0.tsv for both) are created.

What's the difference btw data/airbnb-{train|test}-indoor-filtered.tsv and data/bnb-dataset-*.tsv?

I counted number of lines of some files. airbnb-train-indoor-filtered.tsv has 1565820 lines, bnb-dataset-raw.tsv has 1451945 lines.

Do they have common pictures? Are those filtered by indoor or not? Do they only have noun-filtered phrases?

I got the following error when I run python cities.py --output data/cities.txt. Could you help me with this? Thanks.

selenium.common.exceptions.WebDriverException: Message: Reached error page: about:neterror?e=netTimeout&u=https%3A//en.wikipedia.org/wiki/List_of_the_most_common_U.S._place_names&c=UTF-8&d=The%20server%20at%20en.wikipedia.org%20is%20taking%20too%20long%20to%20respond.

Where is the file 'data/bnb-test-indoor.tsv'?

There are only 2 positional arguments in save_txt(), why do you input 3?

BTW, are you sure save_txt() saves any txt files? Because I only see the print operation there.

Thanks for your reply!

Hi, Thanks for your work to make such a large indoor data set.
However, I don't know whether my computer has the enough memory to save it. What is the total size of this data set？

When I followed your instructions "To parallelize among multiple GPUs, you can use the distributed.launcher script as is...", some problems occurred as follows:

And I found that this warning occurred at the beginning:

How can I fix this problem?

I can't find how to generate places365.hierarchy.txt in any part of the code. How do I get it?

At cities.py, you should include import json.

When I run this program, it shows that the "data/task/noun_phrases.txt" file is missing.
Could you upload this file?
Thanks.

Hi,
I have following error when I try to run
python -m torch.distributed.launch \ --nproc_per_node=24 \ --nnodes=1 \ --node_rank=0 \ -m extract_noun_phrases \ --start 24 \ --num_splits 48 \ --source data/bnb-test-indoor.tsv \ --output noun_phrases.tsv \ --num_workers 8 \ --batch_size 20

from README.md.
Error: unrecognized arguments: –local_rank=19

A quick fix is here: https://discuss.pytorch.org/t/error-unrecognized-arguments-local-rank-1/83679

Could you please amend the extract_noun_phrases.py file?
And I also wonder why the file with same name is in the scripts folder too.

Thank you :)

When I run python search_listings.py --locations data/cities.txt --output data/listings, it may happen the following error:

Traceback (most recent call last):
File "C:\Users\Sznas\AppData\Local\Programs\Python\Python38\lib\multiprocessing\pool.py", line 125, in worker
result = (True, func(*args, **kwds))
File "D:\Pycharm\bnb-dataset-main\search_listings.py", line 113, in search_locations
search_location(row["name"], Path(row["dest"]), limit)
File "D:\Pycharm\bnb-dataset-main\search_listings.py", line 93, in search_location
assert pagination["pageLimit"] == limit, pagination
AssertionError: {'__typename': 'DoraExploreV3PaginationMetadata', 'hasNextPage': True, 'itemsOffset': 40, 'sectionOffset': 3, 'hasPreviousPage': False, 'previousPageSectionOffset': None, 'previousPageItemsOffset': None, 'searchSessionId': 'dd37d185-0c70-4184-9d08-0e8033aa40f5', 'pageLimit': 40, 'totalCount': '114'}
"""

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "search_listings.py", line 175, in
run_downloader(args)
File "search_listings.py", line 156, in run_downloader
list(
File "C:\Users\Sznas\AppData\Local\pypoetry\Cache\virtualenvs\bnb-dataset-sISfIGJa-py3.8\lib\site-packages\tqdm\std.py", line 1195, in iter
for obj in iterable:
File "C:\Users\Sznas\AppData\Local\Programs\Python\Python38\lib\multiprocessing\pool.py", line 868, in next
raise value
AssertionError: {'__typename': 'DoraExploreV3PaginationMetadata', 'hasNextPage': True, 'itemsOffset': 40, 'sectionOffset': 3, 'hasPreviousPage': False, 'previousPageSectionOffset': None, 'previousPageItemsOffset': None, 'searchSessionId': 'dd37d185-0c70-4184-9d08-0e8033aa40f5', 'pageLimit': 40, 'totalCount': '114'}

I download the airbnb-test-indoor-filtered.tsv and airbnb-train-indoor-filtered.tsv，and find that the numbers of images are 118571 and 1523116 respectively, the sum of which is 1687452 and is not equal to “713K image-caption pairs and 676K images without captions” (13000+676000=1389000) in the paper. Could you explain this? Thanks.