The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
text: list<item: string>
child 0, item: string
site: string
server: string
files: list<item: struct<file: string, size: int64, sha256: string, what: string, used_by: string>>
child 0, item: struct<file: string, size: int64, sha256: string, what: string, used_by: string>
child 0, file: string
child 1, size: int64
child 2, sha256: string
child 3, what: string
child 4, used_by: string
to
{'site': Value('string'), 'server': Value('string'), 'files': List({'file': Value('string'), 'size': Value('int64'), 'sha256': Value('string'), 'what': Value('string'), 'used_by': Value('string')})}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 544, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3052, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 558, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 361, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
text: list<item: string>
child 0, item: string
site: string
server: string
files: list<item: struct<file: string, size: int64, sha256: string, what: string, used_by: string>>
child 0, item: struct<file: string, size: int64, sha256: string, what: string, used_by: string>
child 0, file: string
child 1, size: int64
child 2, sha256: string
child 3, what: string
child 4, used_by: string
to
{'site': Value('string'), 'server': Value('string'), 'files': List({'file': Value('string'), 'size': Value('int64'), 'sha256': Value('string'), 'what': Value('string'), 'used_by': Value('string')})}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
websandbox site data
Data for websandbox: lightweight, self-hostable websites for training and evaluating web agents. Each site is one SQLite database (plus media) served by one small Node server; it boots in under a second and resets to its initial state in milliseconds, so one machine can run dozens of independent copies of every site.
The sites are rebuilt from the WebArena environments (same kinds of features and the same underlying data, without Magento, Postmill, GitLab, Kiwix or the OpenStreetMap stack). The WebArena tasks converted to these sites, the generated tasks and the servers are in the GitHub repository; this dataset holds only the data the servers load.
Contents
About 21 GB in total. Each folder is the --data directory of one server
(sites/<site>/server.js --data data/<site>), with a manifest.json (file,
size, sha256, what it is, which server option reads it), SHA256SUMS and
SOURCES.md (origin, licence, attribution).
| Site | Files | Size | What it contains |
|---|---|---|---|
store/ |
base.db, media.tar |
1.16 GB + 4.50 GB | Shopping storefront and its back office (/admin): the WebArena "One Stop Market" (Amazon-derived catalog) merged with the Magento Luma admin store. 106,408 products, 309,290 reviews, 497 orders, 72 customers, CMS pages, price rules, full-text indexes; 195k product images (WebP, long side <= 600 px). |
forum/ |
base.db, media.tar |
1.34 GB + 1.97 GB | Reddit-like forum (Postmill in WebArena): 95 forums, 661,782 users, 127,391 submissions, 2,551,513 comments; 63k submission images and thumbnails (WebP). |
code/ |
base.db, media.tar, SOURCES.repos.md |
2.90 GB | GitLab-like code hosting: 175 projects with git history (768,026 commits; trees/blobs of older commits of large repos omitted), 2,399 users, 80,962 issues, 134,336 merge requests, 303,407 comments, labels, milestones, events, to-dos. |
wiki/ |
base.db, media.tar |
2.93 GB + 0.98 GB | English Wikipedia subset (2022-05-01 Kiwix snapshot): 115,040 articles (all task articles plus the most linked ones), redirects, link graph, full-text search, title suggestions; 146,917 thumbnails (<= 320 px). |
map/ |
geocode.db, tiles/map.mbtiles, route/*, static.tar.zst |
1.71 + 1.53 + 2.06 + 0.05 GB | OpenStreetMap-based map of US Northeast, Greater London, San Francisco and Singapore: vector tiles (OpenMapTiles schema, z0-14), geocoder (2.57M places, 12.3M house numbers), car/bike/foot routing graph, style, glyphs and sprites. |
Download and run
git clone https://github.com/junfengzh/websandbox && cd websandbox
huggingface-cli download junfengzhou/websandbox --repo-type dataset --local-dir data
# unpack the media / static archives in place
for s in store forum code wiki; do tar -xf data/$s/media.tar -C data/$s; done
zstd -dc data/map/static.tar.zst | tar -x -C data/map
(cd data/store && sha256sum -c SHA256SUMS) # optional, per site
# install the Node (>= 20) dependencies once
(cd kit && npm install); for s in forum code wiki map; do (cd sites/$s && npm install); done
tasks/tools/sites.sh up --data-root data # all five sites; `sites.sh urls` prints them
tasks/tools/sites.sh down
The tar files can be deleted after unpacking. A single site:
node sites/forum/server.js --tenants 1 --port 7200 --data data/forum.
Licences and attribution
This dataset combines material under different licences; each site's
SOURCES.md has the details. Use it in accordance with all of them.
websandbox code is Apache-2.0 (GitHub repository).
store/ -- Derived from WebArena (Zhou et al., 2023, https://webarena.dev, Apache-2.0), which includes Magento 2 Luma sample data (OSL-3.0 / AFL-3.0) and product metadata and reviews from the Amazon review data (McAuley et al., UCSD); product texts, review texts and images remain under their original terms. Luma product images: Magento 2 sample data (https://github.com/magento/magento2-sample-data, AFL-3.0 / OSL-3.0, Copyright Magento, Inc. / Adobe). Customer names, e-mail addresses, phone numbers and addresses are synthetic (generated by the WebArena authors).
forum/ -- Derived from WebArena (Apache-2.0), which contains public Reddit posts, comments and images (2022-10 to 2023-03): copyright their authors, subject to Reddit's terms (User Agreement / Data API terms). Items keep their original Reddit ids so they can be traced to their origin. Forum structure modelled on Postmill (https://postmill.xyz, Zlib licence); no Postmill code. IP addresses and e-mail addresses were not exported.
code/ -- Derived from the WebArena GitLab environment, which mirrors public
GitHub repositories with their issues, pull requests and comments. Each
repository remains under its own licence: see
code/SOURCES.repos.md (175 projects, upstream URL and
licence file found). Repositories without a licence file are included only as
part of this research benchmark, as WebArena published them. Issue and comment
text belongs to its GitHub authors (GitHub Terms of Service). Project templates
are from GitLab (https://gitlab.com/gitlab-org/project-templates), each under
its own licence. Access tokens, SSH keys, OAuth/CI/runner tokens, OTP secrets
and webhooks were dropped; the only token kept is the task user's RSS feed
token, which a WebArena task asks for and which is published in WebArena's
task file.
wiki/ -- Text: Wikipedia contributors, Wikipedia, The Free Encyclopedia,
English edition, snapshot of 2022-05-01, via the Kiwix ZIM
wikipedia_en_all_maxi_2022-05 (https://www.kiwix.org/), licensed under
CC BY-SA 4.0; the modified
articles (subset, cleaned HTML, links outside the subset removed) are
distributed under CC BY-SA 4.0 as well. Each article keeps its revision id
(https://en.wikipedia.org/w/index.php?oldid=<oldid> gives its history and
authors). Images: Wikimedia Commons / Wikipedia contributors, under their
individual licences (see each file page).
map/ -- © OpenStreetMap contributors (https://www.openstreetmap.org/copyright), © OpenMapTiles. The tiles, geocoder database and routing graph are derived from OpenStreetMap extracts (Geofabrik: US Northeast, Greater London; BBBike: San Francisco, Singapore) and are available under the Open Database License 1.0. Tiles built with planetiler (Apache-2.0) using the OpenMapTiles schema (BSD-3-Clause code, CC-BY 4.0 design); OSM water polygons (ODbL), Natural Earth (public domain), lake centerlines (MIT, from OSM). Style: OSM Bright GL (BSD-3-Clause / CC-BY 4.0, © OpenMapTiles); sprites derived from Maki (CC0). Glyphs: Noto Sans (SIL Open Font License 1.1, © Google) via openmaptiles/fonts.
Redactions: on top of the WebArena public images, store/base.db has
orders.remote_ip set to NULL for all imported orders, and code/base.db has
live-looking credentials pasted into issue/MR/comment text (GitHub tokens,
Google API keys, AWS keys, Slack tokens, private key bodies) replaced with
[REDACTED] and the personal webmail addresses in users.email replaced with
<username>@example.com. Git data (commit author/committer emails, blobs,
commit messages) is kept as recorded in the public repositories.
Removal requests: if you are the author of content in this dataset and want it removed, open a discussion on this dataset page.
Citation
Please cite WebArena, whose environments this data is derived from:
@inproceedings{zhou2024webarena,
title = {WebArena: A Realistic Web Environment for Building Autonomous Agents},
author = {Zhou, Shuyan and Xu, Frank F. and Zhu, Hao and Zhou, Xuhui and Lo, Robert and Sridhar, Abishek and Cheng, Xianyi and Ou, Tianyue and Bisk, Yonatan and Fried, Daniel and Alon, Uri and Neubig, Graham},
booktitle = {The Twelfth International Conference on Learning Representations (ICLR)},
year = {2024},
url = {https://webarena.dev}
}
- Downloads last month
- 137