Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
images
images list
messages
string
metadata
string
[{"role": "user", "content": [{"type": "text", "index": null, "text": "From a terminal, pull the state name (first column) from the first 5 data rows of ~/Desktop/us-population-states.csv, skipping the header row. Save the five names, one per line, to ~/Desktop/first5_states.txt."}, {"type": "image", "index": 1.0, "tex...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_awk_state_names", "env_id": "lite.osworld", "task_id": "synth_apps_asset_awk_state_names_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "conf...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "In a terminal, combine the four FRED CSVs in ~/Desktop/fred/ (us-fed-funds-rate.csv, us-mortgage-30yr.csv, us-housing-starts.csv, oil-wti-daily.csv) into a single ~/Desktop/all_indicators.csv with header `date,series,value`. For each input file, dro...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_concat_fred_csvs", "env_id": "lite.osworld", "task_id": "synth_apps_asset_concat_fred_csvs_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "co...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Convert ~/Desktop/us-state-median-income.csv into an xlsx workbook at ~/Desktop/us-state-median-income.xlsx using LibreOffice Calc."}, {"type": "image", "index": 1.0, "text": null}], "reasoning_content": null, "tool_calls": null}, {"role": "assistan...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_csv_to_xlsx_state_income", "env_id": "lite.osworld", "task_id": "synth_apps_asset_csv_to_xlsx_state_income_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id...
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open ~/Desktop/us-state-median-income.csv in Calc and save it as ~/Desktop/us-state-median-income.xlsx (Excel xlsx format)."}], "reasoning_content": null, "tool_calls": null}, {"role": "assistant", "con...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_csv_to_xlsx_state_income", "env_id": "lite.osworld", "task_id": "synth_apps_asset_csv_to_xlsx_state_income_0002", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "In a terminal, find the three largest .txt files inside ~/Desktop/gutenberg/ by byte size (smallest first, largest last) and save the result to ~/Desktop/top3_books.txt. Each output line should be the byte count, a tab, and the file path (the same f...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_du_top3_gutenberg", "env_id": "lite.osworld", "task_id": "synth_apps_asset_du_top3_gutenberg_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "From a terminal, filter ~/Desktop/world-gdp-2022.csv down to only the G7 country rows \u2014 those whose CountryCode (first column) is one of CAN, FRA, DEU, ITA, JPN, GBR, or USA. Save the matching rows in their original order to ~/Desktop/g7-gdp.cs...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_grep_g7_world_gdp", "env_id": "lite.osworld", "task_id": "synth_apps_asset_grep_g7_world_gdp_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "In a terminal, save the first 51 lines of ~/Desktop/oil-wti-daily.csv (header + first 50 data rows) to ~/Desktop/oil-wti-head.csv. Use `head -n 51`."}, {"type": "image", "index": 1.0, "text": null}], "reasoning_content": null, "tool_calls": null}, {...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_head_oil_wti", "env_id": "lite.osworld", "task_id": "synth_apps_asset_head_oil_wti_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "config_pat...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "From a terminal, drop every line that contains the word ALGERNON from ~/Desktop/importance-of-being-earnest.txt (leave all other lines untouched) and write the filtered text to ~/Desktop/earnest_no_algernon.txt."}, {"type": "image", "index": 1.0, "t...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_sed_drop_algernon", "env_id": "lite.osworld", "task_id": "synth_apps_asset_sed_drop_algernon_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "From a terminal, pull the first 10 data rows out of ~/Desktop/oil-wti-daily.csv (skip the header row entirely) and save just those 10 rows to ~/Desktop/oil-wti-sample.csv."}, {"type": "image", "index": 1.0, "text": null}], "reasoning_content": null,...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_shuf_oil_wti", "env_id": "lite.osworld", "task_id": "synth_apps_asset_shuf_oil_wti_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "config_pat...
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "In a terminal, sort ~/Desktop/importance-of-being-earnest.txt case-insensitively (use `LC_ALL=C sort -f`) and save the sorted lines to ~/Desktop/earnest-sorted.txt."}], "reasoning_content": null, "tool_...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_sort_case_earnest", "env_id": "lite.osworld", "task_id": "synth_apps_asset_sort_case_earnest_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "In a terminal, sort the lines of ~/Desktop/art-of-war.txt and remove duplicates with `LC_ALL=C sort -u`. Save the deduplicated, sorted result to ~/Desktop/art-of-war-uniq.txt."}, {"type": "image", "index": 1.0, "text": null}], "reasoning_content": n...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:asset_sort_uniq_artofwar", "env_id": "lite.osworld", "task_id": "synth_apps_asset_sort_uniq_artofwar_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua",...
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open ~/Desktop/us-gdp.csv in Calc and save it as ~/Desktop/us-gdp.xlsx."}], "reasoning_content": null, "tool_calls": null}, {"role": "assistant", "content": [{"type": "action_description", "index": null...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:csv_to_xlsx_us_gdp", "env_id": "lite.osworld", "task_id": "synth_apps_csv_to_xlsx_us_gdp_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "config_pat...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Open ~/Desktop/us-population-states.csv in Calc and save it as ~/Desktop/us-population-states.xlsx (Excel xlsx format)."}, {"type": "image", "index": 1.0, "text": null}], "reasoning_content": null, "tool_calls": null}, {"role": "assistant", "content...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:csv_to_xlsx_us_population", "env_id": "lite.osworld", "task_id": "synth_apps_csv_to_xlsx_us_population_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua...
[{"role": "user", "content": [{"type": "image", "index": 0.0, "text": null}, {"type": "text", "index": null, "text": "Open ~/Desktop/us-unemployment.csv in Calc and save it as ~/Desktop/us-unemployment.xlsx."}], "reasoning_content": null, "tool_calls": null}, {"role": "assistant", "content": [{"type": "action_descripti...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:csv_to_xlsx_us_unemployment", "env_id": "lite.osworld", "task_id": "synth_apps_csv_to_xlsx_us_unemployment_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "ev...
[{"role": "user", "content": [{"type": "text", "index": null, "text": "Open ~/Desktop/world-gdp-2022.csv in Calc and save it as ~/Desktop/world-gdp-2022.xlsx."}, {"type": "image", "index": 1.0, "text": null}], "reasoning_content": null, "tool_calls": null}, {"role": "assistant", "content": [{"type": "action_description...
{"platform": "desktop", "task_type": "use", "extra_tool_schemas": [], "valid_actions": null, "others": {"domain": "multi_apps", "source": "synthetic:csv_to_xlsx_world_gdp", "env_id": "lite.osworld", "task_id": "synth_apps_csv_to_xlsx_world_gdp_0001", "model_id": "meituan/EvoCUA-8B-20260105", "agent_id": "evocua", "conf...
End of preview. Expand in Data Studio

cua-lite/LiteOS

LiteOS desktop-use agent rollout dataset. Train split is preserved from HaoranLiu/LiteOS; validation split is staged from .logs/rollout/lite.osworld_eval_g3 with model folder names matching train split configs.

Origin

Load via datasets

from datasets import load_dataset

# entire dataset
ds = load_dataset("cua-lite/LiteOS")

# just one named subset (config)
ds = load_dataset("cua-lite/LiteOS", "EvoCUA-8B-20260105")

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 the configs: 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 (the datasets library blacklists <>:/\|?* in split names; everything else is fine in config_name)
  • validation is an in-distribution held-out slice (never used in training); test is reserved for out-of-distribution benchmark datasets

Stats

platform task_type variant train validation
desktop use EvoCUA-8B-20260105 1,703 996
desktop use Qwen3-VL-2B-Instruct 1,697 996
desktop use Qwen3-VL-32B-Instruct 1,702 996
desktop use Qwen3-VL-4B-Instruct 1,703 996
desktop use Qwen3-VL-8B-Instruct 1,704 996
desktop use Qwen3.5-27B 5,112 996
desktop use Qwen3.5-2B 1,701 995
desktop use Qwen3.5-4B 1,704 995
desktop use Qwen3.5-9B 1,704 996
desktop use Qwen3.8-27B 5,046 994
desktop use ScaleCUA-7B 1,676 0
desktop use UI-TARS-1.5-7B 1,704 996
desktop use UI-TARS-7B-DPO 1,704 993
desktop use gpt-5.4 1,704 996
desktop use gpt-5.5 1,702 996
desktop use gpt-5.6-sol 0 994
desktop use gpt-6-astra 0 993

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/LiteOS/
  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: 632,354
  • Image store size: 285.91 GB

Notes

Staged via lite.data.hf.stage from rollout log-roots: .logs/rollout/lite.osworld_eval_g3/Qwen3-VL-2B-Instruct, .logs/rollout/lite.osworld_eval_g3/Qwen3-VL-32B-Instruct, .logs/rollout/lite.osworld_eval_g3/Qwen3-VL-4B-Instruct, .logs/rollout/lite.osworld_eval_g3/Qwen3-VL-8B-Instruct, .logs/rollout/lite.osworld_eval_g3/Qwen3.5-27B, .logs/rollout/lite.osworld_eval_g3/Qwen3.5-2B, .logs/rollout/lite.osworld_eval_g3/Qwen3.5-4B, .logs/rollout/lite.osworld_eval_g3/Qwen3.5-9B, .logs/rollout/lite.osworld_eval_g3/Qwen3.8-27B, .logs/rollout/lite.osworld_eval_g3/UI-TARS-1.5-7B, .logs/rollout/lite.osworld_eval_g3/UI-TARS-7B-DPO, .logs/rollout/lite.osworld_eval_g3/evocua_8b, .logs/rollout/lite.osworld_eval_g3/gpt-5.4, .logs/rollout/lite.osworld_eval_g3/gpt-5.5 (row filter: none).

License & citation

other

Downloads last month
34