Dataset Viewer
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
agent_id: int64
interaction_id: int64
prompt: string
request_messages: list<item: struct<role: string, content: string>>
child 0, item: struct<role: string, content: string>
child 0, role: string
child 1, content: string
memory_context: string
memory_screenshot_steps: list<item: int64>
child 0, item: int64
model_response_text: string
model_reasoning_text: string
model_tool_calls: list<item: struct<name: string, arguments: string>>
child 0, item: struct<name: string, arguments: string>
child 0, name: string
child 1, arguments: string
parsed_action: struct<action: string, key: string, reason: string>
child 0, action: string
child 1, key: string
child 2, reason: string
action_valid: bool
action_validity_reason: string
executed_action: struct<action: string, key: string, reason: string, duration: double>
child 0, action: string
child 1, key: string
child 2, reason: string
child 3, duration: double
game_state: struct<schemaVersion: string, gameId: string, seed: int64, timestampMs: double, gameTimeMs: double, (... 981 chars omitted)
child 0, schemaVersion: string
child 1, gameId: string
child 2, seed: int64
child 3, timestampMs: double
child 4, gameTimeMs: double
child 5, status: string
child 6, is_actionable: bool
child 7, terminal: struct<isTerminal: bool, outcome: string, reason: string>
child 0, isTerminal: bool
child 1, outcome: string
child 2, reason: string
child 8, game_state: struct<score: int
...
9, metrics.distance: int64
child 10, metrics.platforms_visible: int64
child 11, terminal.outcome: string
child 12, terminal.isTerminal: bool
screenshot: string
interaction_count: int64
final_metrics: struct<objective_target_run_bests: struct<game_state.score: double>, progress: double, score: double (... 434 chars omitted)
child 0, objective_target_run_bests: struct<game_state.score: double>
child 0, game_state.score: double
child 1, progress: double
child 2, score: double
child 3, gameTimeMs: double
child 4, metrics.distance: int64
child 5, metrics.platforms_visible: int64
child 6, terminal.outcome: null
child 7, terminal.isTerminal: bool
child 8, objective_target_bests: struct<game_state.score: double>
child 0, game_state.score: double
child 9, objective_target_results: list<item: struct<field: string, start: double, target: double, actual: int64, best: double, run_bes (... 40 chars omitted)
child 0, item: struct<field: string, start: double, target: double, actual: int64, best: double, run_best: double, (... 28 chars omitted)
child 0, field: string
child 1, start: double
child 2, target: double
child 3, actual: int64
child 4, best: double
child 5, run_best: double
child 6, progress: double
child 7, met: bool
child 10, objective_targets_met: bool
child 11, progress_current: double
child 12, target_reached: bool
agent_count: int64
outcome: string
to
{'agent_count': Value('int64'), 'interaction_count': Value('int64'), 'outcome': Value('string'), 'final_metrics': {'objective_target_run_bests': {'game_state.score': Value('float64')}, 'progress': Value('float64'), 'score': Value('float64'), 'gameTimeMs': Value('float64'), 'metrics.distance': Value('int64'), 'metrics.platforms_visible': Value('int64'), 'terminal.outcome': Value('null'), 'terminal.isTerminal': Value('bool'), 'objective_target_bests': {'game_state.score': Value('float64')}, 'objective_target_results': List({'field': Value('string'), 'start': Value('float64'), 'target': Value('float64'), 'actual': Value('int64'), 'best': Value('float64'), 'run_best': Value('float64'), 'progress': Value('float64'), 'met': Value('bool')}), 'objective_targets_met': Value('bool'), 'progress_current': Value('float64'), 'target_reached': Value('bool')}}
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
agent_id: int64
interaction_id: int64
prompt: string
request_messages: list<item: struct<role: string, content: string>>
child 0, item: struct<role: string, content: string>
child 0, role: string
child 1, content: string
memory_context: string
memory_screenshot_steps: list<item: int64>
child 0, item: int64
model_response_text: string
model_reasoning_text: string
model_tool_calls: list<item: struct<name: string, arguments: string>>
child 0, item: struct<name: string, arguments: string>
child 0, name: string
child 1, arguments: string
parsed_action: struct<action: string, key: string, reason: string>
child 0, action: string
child 1, key: string
child 2, reason: string
action_valid: bool
action_validity_reason: string
executed_action: struct<action: string, key: string, reason: string, duration: double>
child 0, action: string
child 1, key: string
child 2, reason: string
child 3, duration: double
game_state: struct<schemaVersion: string, gameId: string, seed: int64, timestampMs: double, gameTimeMs: double, (... 981 chars omitted)
child 0, schemaVersion: string
child 1, gameId: string
child 2, seed: int64
child 3, timestampMs: double
child 4, gameTimeMs: double
child 5, status: string
child 6, is_actionable: bool
child 7, terminal: struct<isTerminal: bool, outcome: string, reason: string>
child 0, isTerminal: bool
child 1, outcome: string
child 2, reason: string
child 8, game_state: struct<score: int
...
9, metrics.distance: int64
child 10, metrics.platforms_visible: int64
child 11, terminal.outcome: string
child 12, terminal.isTerminal: bool
screenshot: string
interaction_count: int64
final_metrics: struct<objective_target_run_bests: struct<game_state.score: double>, progress: double, score: double (... 434 chars omitted)
child 0, objective_target_run_bests: struct<game_state.score: double>
child 0, game_state.score: double
child 1, progress: double
child 2, score: double
child 3, gameTimeMs: double
child 4, metrics.distance: int64
child 5, metrics.platforms_visible: int64
child 6, terminal.outcome: null
child 7, terminal.isTerminal: bool
child 8, objective_target_bests: struct<game_state.score: double>
child 0, game_state.score: double
child 9, objective_target_results: list<item: struct<field: string, start: double, target: double, actual: int64, best: double, run_bes (... 40 chars omitted)
child 0, item: struct<field: string, start: double, target: double, actual: int64, best: double, run_best: double, (... 28 chars omitted)
child 0, field: string
child 1, start: double
child 2, target: double
child 3, actual: int64
child 4, best: double
child 5, run_best: double
child 6, progress: double
child 7, met: bool
child 10, objective_targets_met: bool
child 11, progress_current: double
child 12, target_reached: bool
agent_count: int64
outcome: string
to
{'agent_count': Value('int64'), 'interaction_count': Value('int64'), 'outcome': Value('string'), 'final_metrics': {'objective_target_run_bests': {'game_state.score': Value('float64')}, 'progress': Value('float64'), 'score': Value('float64'), 'gameTimeMs': Value('float64'), 'metrics.distance': Value('int64'), 'metrics.platforms_visible': Value('int64'), 'terminal.outcome': Value('null'), 'terminal.isTerminal': Value('bool'), 'objective_target_bests': {'game_state.score': Value('float64')}, 'objective_target_results': List({'field': Value('string'), 'start': Value('float64'), 'target': Value('float64'), 'actual': Value('int64'), 'best': Value('float64'), 'run_best': Value('float64'), 'progress': Value('float64'), 'met': Value('bool')}), 'objective_targets_met': Value('bool'), 'progress_current': Value('float64'), 'target_reached': Value('bool')}}
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.
GameWorld Benchmark Trajectories
These trajectories were collected using the GameWorld benchmark.
This dataset is organized for inspecting trajectories. It contains 1,020 completed trajectories from Qwen3.8-Flash-Next, Gemini-3.8-Flash, and GPT-6.1-Terra, using both General and Computer-Use (CUA) interfaces on the GameWorld 400-step benchmark: 170 tasks and one run per task for each model and interface.
Leaderboard
Each model was evaluated on 170 tasks with a 400-step limit per run. Qwen scores average all three runs per task; Gemini and GPT scores use one run per task. SR is the success rate, and PG is mean task progress.
| Model | Runs/task | CUA SR | CUA PG | General SR | General PG |
|---|---|---|---|---|---|
| Qwen3.5-9B | 3 | 14.51% | 35.09% | 17.65% | 39.32% |
| Qwen3.5-27B | 3 | 21.96% | 41.79% | 16.27% | 39.59% |
| Qwen3.5-35B-A3B | 3 | 22.94% | 43.82% | 19.41% | 39.38% |
| Qwen3.5-122B-A10B | 3 | 23.14% | 45.23% | 19.02% | 41.87% |
| Qwen3.6-27B | 3 | 21.18% | 43.60% | 18.82% | 43.25% |
| Qwen3.6-35B-A3B | 3 | 19.22% | 40.57% | 15.69% | 38.84% |
| Qwen3.8-27B | 3 | 29.22% | 54.64% | 24.12% | 49.34% |
| Qwen3.8-Flash-Next | 3 | 36.86% | 61.48% | 29.80% | 55.86% |
| Gemini-3.8-Flash | 1 | 36.47% | 60.11% | 42.35% | 63.52% |
| GPT-6.1-Terra | 1 | 55.29% | 72.53% | 61.18% | 76.79% |
The trajectory archives contain one run per model and interface; the Qwen leaderboard scores above aggregate the original three runs.
Files
README.md
manifest.json
qwen3.8-flash-next_general.zip
qwen3.8-flash-next_cua.zip
gemini-3.8-flash_general.zip
gemini-3.8-flash_cua.zip
gpt-6.1-terra_general.zip
gpt-6.1-terra_cua.zip
examples/
qwen3.8-flash-next-general-doodle-jump/
qwen3.8-flash-next-cua-doodle-jump/
gemini-3.8-flash-general-doodle-jump/
gemini-3.8-flash-cua-doodle-jump/
gpt-6.1-terra-general-doodle-jump/
gpt-6.1-terra-cua-doodle-jump/
Each ZIP contains 170 trajectories from one model and one interface. After extracting, the folders look like:
qwen3.8-flash-next_general/
10_doodle-jump/
10_01/
episode.json
steps.jsonl
screenshots/a0_001.png
screenshots/a0_002.png
...
The game_id/task_id directory identifies a trajectory.
episode.json contains its agent and interaction counts, outcome, and final metrics.
Each line of steps.jsonl describes one interaction and gives a screenshot path
relative to that trajectory's directory.
Examples provide the same records for inspecting individual steps:
examples/qwen3.8-flash-next-general-doodle-jump/
episode.json
steps.jsonl
steps/001.json
steps/001.png
...
The benchmark protocol is fine-control-2026-08 with a maximum of 400 steps
per trajectory. See the GameWorld GitHub repository
for the benchmark code, task definitions, and evaluation design.
- Downloads last month
- 562