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
action: struct<additionalProperties: bool, description: string, properties: struct<metadata: struct<addition (... 601 chars omitted)
child 0, additionalProperties: bool
child 1, description: string
child 2, properties: struct<metadata: struct<additionalProperties: bool, description: string, title: string, type: string (... 503 chars omitted)
child 0, metadata: struct<additionalProperties: bool, description: string, title: string, type: string>
child 0, additionalProperties: bool
child 1, description: string
child 2, title: string
child 3, type: string
child 1, action_type: struct<default: string, title: string, type: string>
child 0, default: string
child 1, title: string
child 2, type: string
child 2, command: struct<default: string, title: string, type: string>
child 0, default: string
child 1, title: string
child 2, type: string
child 3, session_id: struct<anyOf: list<item: struct<type: string>>, default: null, title: string>
child 0, anyOf: list<item: struct<type: string>>
child 0, item: struct<type: string>
child 0, type: string
child 1, default: null
child 2, title: string
child 4, block: struct<default: bool, title: string, type: string>
child 0, default: bool
child 1, title: string
child 2, type: string
child 5, wait_seconds: struct<anyOf
...
0, additionalProperties: bool
child 1, title: string
child 2, type: string
child 3, title: string
child 4, type: string
files: struct<arena/concordance_records/__init__.py: string, arena/concordance_records/app.py: string, aren (... 888 chars omitted)
child 0, arena/concordance_records/__init__.py: string
child 1, arena/concordance_records/app.py: string
child 2, arena/concordance_records/catalog.py: string
child 3, arena/concordance_records/cli.py: string
child 4, arena/concordance_records/Dockerfile: string
child 5, arena/concordance_records/families/__init__.py: string
child 6, arena/concordance_records/families/capacity.py: string
child 7, arena/concordance_records/families/contacts.py: string
child 8, arena/concordance_records/families/headcount.py: string
child 9, arena/concordance_records/families/heat.py: string
child 10, arena/concordance_records/families/lab_units.py: string
child 11, arena/concordance_records/families/ledger.py: string
child 12, arena/concordance_records/jail.py: string
child 13, arena/concordance_records/LICENSE-OpenEnv.txt: string
child 14, arena/concordance_records/models.py: string
child 15, arena/concordance_records/runtime.py: string
child 16, arena/concordance_records/verifier.py: string
child 17, LICENSE-OpenEnv.txt: string
child 18, NOTICE.txt: string
child 19, README.md: string
child 20, schema.json: string
child 21, submission.json: string
child 22, tasks.jsonl: string
to
{'files': {'arena/concordance_records/__init__.py': Value('string'), 'arena/concordance_records/app.py': Value('string'), 'arena/concordance_records/catalog.py': Value('string'), 'arena/concordance_records/cli.py': Value('string'), 'arena/concordance_records/Dockerfile': Value('string'), 'arena/concordance_records/families/__init__.py': Value('string'), 'arena/concordance_records/families/capacity.py': Value('string'), 'arena/concordance_records/families/contacts.py': Value('string'), 'arena/concordance_records/families/headcount.py': Value('string'), 'arena/concordance_records/families/heat.py': Value('string'), 'arena/concordance_records/families/lab_units.py': Value('string'), 'arena/concordance_records/families/ledger.py': Value('string'), 'arena/concordance_records/jail.py': Value('string'), 'arena/concordance_records/LICENSE-OpenEnv.txt': Value('string'), 'arena/concordance_records/models.py': Value('string'), 'arena/concordance_records/runtime.py': Value('string'), 'arena/concordance_records/verifier.py': Value('string'), 'LICENSE-OpenEnv.txt': Value('string'), 'NOTICE.txt': Value('string'), 'README.md': Value('string'), 'schema.json': Value('string'), 'submission.json': Value('string'), 'tasks.jsonl': Value('string')}}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, 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 127, 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 483, 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 2951, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, 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 547, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, 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 343, 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 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
action: struct<additionalProperties: bool, description: string, properties: struct<metadata: struct<addition (... 601 chars omitted)
child 0, additionalProperties: bool
child 1, description: string
child 2, properties: struct<metadata: struct<additionalProperties: bool, description: string, title: string, type: string (... 503 chars omitted)
child 0, metadata: struct<additionalProperties: bool, description: string, title: string, type: string>
child 0, additionalProperties: bool
child 1, description: string
child 2, title: string
child 3, type: string
child 1, action_type: struct<default: string, title: string, type: string>
child 0, default: string
child 1, title: string
child 2, type: string
child 2, command: struct<default: string, title: string, type: string>
child 0, default: string
child 1, title: string
child 2, type: string
child 3, session_id: struct<anyOf: list<item: struct<type: string>>, default: null, title: string>
child 0, anyOf: list<item: struct<type: string>>
child 0, item: struct<type: string>
child 0, type: string
child 1, default: null
child 2, title: string
child 4, block: struct<default: bool, title: string, type: string>
child 0, default: bool
child 1, title: string
child 2, type: string
child 5, wait_seconds: struct<anyOf
...
0, additionalProperties: bool
child 1, title: string
child 2, type: string
child 3, title: string
child 4, type: string
files: struct<arena/concordance_records/__init__.py: string, arena/concordance_records/app.py: string, aren (... 888 chars omitted)
child 0, arena/concordance_records/__init__.py: string
child 1, arena/concordance_records/app.py: string
child 2, arena/concordance_records/catalog.py: string
child 3, arena/concordance_records/cli.py: string
child 4, arena/concordance_records/Dockerfile: string
child 5, arena/concordance_records/families/__init__.py: string
child 6, arena/concordance_records/families/capacity.py: string
child 7, arena/concordance_records/families/contacts.py: string
child 8, arena/concordance_records/families/headcount.py: string
child 9, arena/concordance_records/families/heat.py: string
child 10, arena/concordance_records/families/lab_units.py: string
child 11, arena/concordance_records/families/ledger.py: string
child 12, arena/concordance_records/jail.py: string
child 13, arena/concordance_records/LICENSE-OpenEnv.txt: string
child 14, arena/concordance_records/models.py: string
child 15, arena/concordance_records/runtime.py: string
child 16, arena/concordance_records/verifier.py: string
child 17, LICENSE-OpenEnv.txt: string
child 18, NOTICE.txt: string
child 19, README.md: string
child 20, schema.json: string
child 21, submission.json: string
child 22, tasks.jsonl: string
to
{'files': {'arena/concordance_records/__init__.py': Value('string'), 'arena/concordance_records/app.py': Value('string'), 'arena/concordance_records/catalog.py': Value('string'), 'arena/concordance_records/cli.py': Value('string'), 'arena/concordance_records/Dockerfile': Value('string'), 'arena/concordance_records/families/__init__.py': Value('string'), 'arena/concordance_records/families/capacity.py': Value('string'), 'arena/concordance_records/families/contacts.py': Value('string'), 'arena/concordance_records/families/headcount.py': Value('string'), 'arena/concordance_records/families/heat.py': Value('string'), 'arena/concordance_records/families/lab_units.py': Value('string'), 'arena/concordance_records/families/ledger.py': Value('string'), 'arena/concordance_records/jail.py': Value('string'), 'arena/concordance_records/LICENSE-OpenEnv.txt': Value('string'), 'arena/concordance_records/models.py': Value('string'), 'arena/concordance_records/runtime.py': Value('string'), 'arena/concordance_records/verifier.py': Value('string'), 'LICENSE-OpenEnv.txt': Value('string'), 'NOTICE.txt': Value('string'), 'README.md': Value('string'), 'schema.json': Value('string'), 'submission.json': Value('string'), 'tasks.jsonl': 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.
Concordance Records
Twelve terminal tasks for reconciling records, across six original families. Each family has two versions. All records are simulated; no peer-authored datasets or real personal or clinical records are included.
| Family | Task | Added rule in the second version |
|---|---|---|
| Contacts | Resolve aliases and field updates. | Withdrawals, cutoff and explicit field clearing. |
| Workforce | Reconcile hires, exits and transfers using effective dates. | Event revisions and fractional staffing. |
| Receivables | Reconcile signed invoices and versioned payments. | Currency conversion and split allocations. |
| Capacity | Calculate equipment availability and capacity factors. | Partial months and overlapping outages. |
| Process energy | Reconcile heat measurements and reported totals. | Reporting-period and feedstock exclusions. |
| Instrument results | Normalize units and exclude failed quality controls. | Latest passing rerun and dilution correction. |
Execution and reward
The agent reads ordinary files and writes answer.json in an isolated working
directory. Instructions and rules.md describe the required output and all
scored business rules. Python and shell tools are available. Candidate commands
run in a separate Linux jail; the trusted controller and reference answers stay
outside it. Route all commands through OpenEnv actions, not through docker exec.
The controller freezes the submitted JSON, closes the working jail and compares
typed values with a private reference. Submitted content is never executed by
the scorer. Reward is 1 only when every requirement passes, otherwise 0. Equivalent
record order and numeric encodings are accepted. Infrastructure failures are
errors. evaluate and close both finalize with the same scorer.
First versions allow 4096 completion / 8192 context tokens; second versions allow 8192 / 16384, with 32 actions each. Task-list order does not prescribe an Arena curriculum. Defaults select twelve frozen instances; explicit reset seeds select new instances of those same templates.
Qualification
CPU checks cover 1200 generated worlds with independently implemented public-input solvers, hand-derived fixtures and plausible incorrect methods. All twelve default tasks were solved inside the actual jail. Native tests also cover malformed and forged artifacts, private-path access, concurrent isolation and cleanup. The published image matches the tested image configuration. These are technical tests; model difficulty and learning benefit have not been measured for this candidate.
Run
docker run --rm -p 127.0.0.1:8000:8000 ghcr.io/sagace-labs/sagace-rlenv-math-budget-20261018@sha256:34c791970362ffec3b1a353ae02b29823adee20e154b7ed1644f2743b43b6b54
openenv validate --url http://localhost:8000
/schema describes the OpenEnv interface. The image uses OpenEnv revision
86a180ede21e044f7929b9a7783ad83aa67d83a3. The source Dockerfile is at
arena/concordance_records/Dockerfile, built from this dataset's root directory.
No task downloads are needed at episode runtime. See tasks.jsonl for the data
and submission.json for the exact image and limits. Rights: see NOTICE.txt.
- Downloads last month
- -