Datasets:
images images list | messages string | metadata string |
|---|---|---|
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Set Chrome's default font size to Very Large."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open on a New Tab page, and the ad... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; python3 -c \"import json, os; p='/home/user/chrome-data/Default/Pre... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Change Chrome's default search engine to Bing."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open with the address bar focused... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; sleep 2; python3 - <<'PY'\nimport json, os, pathlib, shutil\npaths ... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Enable 'Do Not Track' in Chrome and set the browser's default font size to Large."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome i... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; sleep 1; python3 - <<'PY'\nimport json, os\npaths = ['/home/user/ch... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open a new tab and navigate to wikipedia.org."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open and I can see the current Air... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": ["google-chrome", "--remote-debugging-port=1337", "--no-default-browser-check", "https://www.wikipedia.org/"], "seconds": null, "window_name": null}, ... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's settings page."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is already open, and the address bar is visible with \... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_06fe7178_4491_4589_810f_2e2bc9502122_traj_verify_1", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's history page."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is already open, and the address bar is visible. To ope... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_06fe7178_4491_4589_810f_2e2bc9502122_traj_verify_2", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's settings page."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is already open, and the address bar is visible at the... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_0d8b7de3_e8de_4d86_b9fd_dd2dce58a217_traj_verify_0", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's browsing history page."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is already open, and the user wants Chrome\u20... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_0d8b7de3_e8de_4d86_b9fd_dd2dce58a217_traj_verify_1", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Change Chrome's default font size to Large."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open and the address bar is visible.... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; sleep 2; python3 - <<'PY'\nimport json, os, pathlib, shutil\npaths ... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Navigate to the Exceptions topic page on this CS50 Python course."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "The file manager window ... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_0e5303d4_8820_42f6_b18d_daf7e633de21_traj_verify_0", "model_id": "gpt-5... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Download the lecture slide PDF for the Conditionals week and save it in the opened lecture_slides folder."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index":... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "oracle_actions": [{"parameters": {"command": "python3 - <<'PY'\nfrom pathlib import Path\nimport os\npath = Path('/home/user/lecture_slides/lecture1.pdf')\nprefix = b'%PDF-1.4\\n% oracle b2\\... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Navigate to the Unit Tests topic page on this CS50 Python course website."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "The browser alre... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_0e5303d4_8820_42f6_b18d_daf7e633de21_traj_verify_2", "model_id": "gpt-5... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open your browsing history in Chrome."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is already open on a New Tab page, and the us... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_12086550_11c0_466b_b367_1d9e75b3910e_traj_verify_0", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's extensions page to review all installed extensions."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open on a new ... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_12086550_11c0_466b_b367_1d9e75b3910e_traj_verify_1", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's bookmark manager."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open on a new tab, and the address bar is alread... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_12086550_11c0_466b_b367_1d9e75b3910e_traj_verify_2", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Navigate to the payment methods section in Chrome's autofill settings."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open on a... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_12086550_11c0_466b_b367_1d9e75b3910e_traj_verify_3", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's built-in password manager to view stored login credentials."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "action_description", "index": null, "text": "Sorry, I can\u... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_12086550_11c0_466b_b367_1d9e75b3910e_traj_verify_4", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Navigate to the password security checkup section in Chrome's password manager."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "The Chrome... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_12086550_11c0_466b_b367_1d9e75b3910e_traj_verify_5", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Navigate to the Dota 2 Official Soundtrack page on Steam Store."}, {"type": "image", "index": 1.0, "text": null}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "The Steam Store home page ... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_121ba48f_9e17_48ce_9bc6_a4fb17a7ebba_traj_verify_0", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Open the Dota 2 Remixes EP DLC page on Steam Store."}, {"type": "image", "index": 1.0, "text": null}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "The Steam Store home page is visible, ... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_121ba48f_9e17_48ce_9bc6_a4fb17a7ebba_traj_verify_1", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Search for 'Dota 2 Soundtrack' on the Steam Store."}, {"type": "image", "index": 1.0, "text": null}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "The Steam Store home page is now visibl... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_121ba48f_9e17_48ce_9bc6_a4fb17a7ebba_traj_verify_2", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's settings page."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is already open, and the address bar is visible at the... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_1704f00f_79e6_43a7_961b_cedd3724d5fd_traj_verify_0", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Change Chrome's default font size to Large."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open and the address bar is visible,... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; python3 - <<'PY'\nimport json, os, time, uuid\nprofile_dirs = ['/ho... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Bookmark the current page in Chrome's bookmarks bar."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open on rentalcars.com, and... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; python3 - <<'PY'\nimport json, os, time, uuid\nprofile_dirs = ['/ho... | |
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Open the ArXiv PDF page for the first paper (StarCoder 2) listed in the document. The ArXiv link is recorded at the end of that paper's entry."}, {"type": "image", "index": 1.0, "text": null}], "tool_calls": null}, {"role": "assistant", "content": [... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "oracle_actions": [{"parameters": {"command": ["google-chrome", "--remote-debugging-port=1337", "--no-default-browser-check", "https://arxiv.org/pdf/2402.19173"], "seconds": null, "window_name... | |
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Add a title 'Paper Reading Notes - March 2024' at the very beginning of the opened document, before any existing content."}, {"type": "image", "index": 1.0, "text": null}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reas... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "oracle_actions": [{"parameters": {"command": "killall -9 -q soffice soffice.bin libreoffice 2>/dev/null || true; python3 - <<'PY'\nimport os\nfrom docx import Document\npath = '/home/user/Des... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Search for the paper 'StarCoder 2 and The Stack v2' on Google using the Chrome browser."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Ch... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_236833a3_5704_47fc_888c_4f298f09f799_traj_verify_2", "model_id": "gpt-5... | |
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Add the Macy's homepage to the bookmarks bar."}, {"type": "image", "index": 1.0, "text": null}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "The page is now at macys.com and shows Macy\... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; python3 - <<'PY'\nimport json, os, time, uuid\nprofile_dirs = ['/ho... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open Chrome's downloads page."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is already open, and the address bar is visible. To o... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "source_domain": "chrome", "source_split": "rl_tasks", "env_id": "lite.scalecua", "task_id": "scalecua_osworld_rl_chrome_2888b4e6_5b47_4b57_8bf5_c73827890774_traj_verify_4", "model_id": "gpt-5.5",... | |
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Create a new folder called 'Work' on the bookmarks bar in Chrome."}], "tool_calls": null}, {"role": "assistant", "content": [{"type": "inline_reasoning", "index": null, "text": "Chrome is open on a new ... | {"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "chrome", "oracle_actions": [{"parameters": {"command": "killall -9 -q chrome google-chrome chromium chromium-browser 2>/dev/null || true; python3 - <<'PY'\nimport json, os, time, uuid\nprofile_dirs = ['/ho... |
cua-lite/Lite.ScaleCUA
Lite.ScaleCUA GPT-5.5 grounded teacher trajectories; every trajectory kept, quality gates tagged in metadata.others.exclude_reason (filter with not exclude_reason and episode_return>0.5)
Origin
Load via datasets
from datasets import load_dataset
# entire dataset
ds = load_dataset("cua-lite/Lite.ScaleCUA")
# just one named subset (config)
ds = load_dataset("cua-lite/Lite.ScaleCUA", "desktop.use.rl")
You can also filter by metadata.platform / metadata.task_type /
metadata.others.* after loading; every row carries a rich metadata
struct (see schema below).
Schema
Each row has these columns:
| column | type | notes |
|---|---|---|
images |
list[Image] | embedded PNG/JPEG bytes; HF viewer renders thumbnails |
messages |
list[struct] | OpenAI-style turns with role + structured content |
metadata |
struct | {platform, task_type, extra_tool_schemas, valid_actions, others{...}} |
Coordinate values in messages are normalized to [0, 1000] integers.
Image-dedup (grounding.* / understanding cohorts). These cohorts are
single-image-per-row and many rows share the same screenshot, so to avoid
re-embedding identical image bytes once per instruction they are stored
folded: one row per unique screenshot (image embedded once), carrying an
extra _folded column — a JSON string with the authoritative list of
{messages, metadata} members for that screenshot. The row's top-level
messages is the members concatenated for viewer convenience. use
cohorts are not folded. Use lite.data.hf.download to consume this repo —
it unfolds automatically back to one row per instruction; reading the parquet
directly yields the folded form.
Layout
<platform>/<task_type>/<split>/shard-NNNNN-of-NNNNN.parquet # single-variant cohort
<platform>/<task_type>/<split>/<variant>/shard-NNNNN-of-NNNNN.parquet # multi-variant cohort
platform∈ {desktop, mobile, web}task_type∈ {understanding, grounding.action, grounding.point, grounding.bbox, use} — used verbatim as the dir component- HF config names are
<platform>.<task_type>by default (e.g.mobile.grounding.action) — UNLESS the dataset was staged with--config-names, which sets verbatim, explicitly-chosen config names (see theconfigs:block above for the authoritative list). The agent registry lookup key in code is<agent>@<platform>@<task_type>(e.g.qwen3_vl@mobile@grounding.action); only this user-facing token uses.between platform and task_type, because@triggers a 403 on the dataset-viewer's signed image URLs. - HF split names stay
train/validation(thedatasetslibrary blacklists<>:/\|?*in split names; everything else is fine in config_name) validationis an in-distribution held-out slice (never used in training);testis reserved for out-of-distribution benchmark datasets
Stats
| platform | task_type | variant | train | validation |
|---|---|---|---|---|
| desktop | use | desktop.use.rl | 1,836 | 0 |
| desktop | use | desktop.use.train | 16,113 | 0 |
Local mirror & SFT export
For local workflows (SFT export, dedup, mixing across datasets), use
lite.data.hf.download to mirror this repo back to the canonical local
layout:
$CUA_LITE_DATASETS_ROOT/cua-lite/Lite.ScaleCUA/
images/<hash[:2]>/<hash>.<ext> # content-addressed image store
<platform>/<task_type>/<split>[/<variant>].parquet # rows reference images by relative path
Rows in the local parquet have images: list[str]; bytes are extracted to
the image store. lite.train.export.export_sft consumes the local
form directly with --image-root=$CUA_LITE_DATASETS_ROOT.
- Total unique images: 243,523
- Image store size: 106.56 GB
Notes
Staged via lite.data.hf.stage from rollout log-roots: .data/rollout/lite.scalecua/gpt/5c3429d9/rl_annotated, .data/rollout/lite.scalecua/gpt/5c3429d9/train_annotated (row filter: none).
License & citation
other
- Downloads last month
- 691