The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 342, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
return check_status(status)
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Column(/families/[]/question/criteria) changed from object to array in row 0
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 101, in _split_generators
pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 304, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
examples = [ujson_loads(line) for line in original_batch.splitlines()]
~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
ValueError: Expected object or value
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Sys1 decision benchmarks
Public synthetic decision cases and recorded adapter studies from Sys1. Use them to inspect how a particular model and adapter answered yes/no, choice, and score questions. The evaluation guide explains the studies and links their original reports.
Sys1 connects coding-agent workflows and application code to Jev, TypeSafe's hosted System One decision model, experimental local Qwen models, or a compatible server. Introducing Sys1 explains the project; the System One guide explains the question types. Sharing an answer format does not imply equal accuracy or calibration across these models.
Files and use
data/benchmarks/ contains the original JSON fixtures and expected answers:
| Fixture | Scope |
|---|---|
forms-v1.json |
Historical 20-case form-action study |
decisions-v2.json |
72 synthetic decisions in nine workflow families |
decisions-v3.json |
72 decisions in nine new families, authored after v2 failures were known |
data/site/data/ contains six complete historical reports: the local and Jev
form studies, Qwen3 0.6B and 1.7B on v2, and Qwen3.5 4B and direct Laya MLX on
v3. The files preserve their original schemas, model responses, failure counts,
timing observations, and source identities. They are JSON documents with nested
records, rather than a shared table. Load each document with a JSON parser.
The v2 and v3 fixture versions contain different tasks. Their scores do not measure a controlled improvement. The v3 cases were frozen before Qwen3.5 4B execution. Historical results describe the named adapter and runtime at the recorded date. Direct Laya is a research adapter outside the shipped backends.
Method and limits
The reproduction guide defines grading, call limits, model pins, and hashes. For v2 and v3, correctness uses the first measured pass. Repeated timing passes and option permutations are observations on the same cases, not additional independent examples. Incorrect answers and failed calls remain in the reports.
Qwen timing includes local worker communication; Jev timing includes HTTP and network time; direct Laya timing excludes both IPC and HTTP. The shared host was not isolated, and repeated requests may benefit from caching. These timing boundaries prevent a direct inference-speed ranking. Token counts also use different tokenizers. Read the report's environment and source fields before comparing observations.
These small, project-authored synthetic studies measure adapter behavior.
They do not establish calibration, production safety, or general model quality.
The separate external JevBench leaderboard is not included in this dataset.
The historical code-review studies in benchmarks/reviewer/,
benchmarks/reviewer-next/, and benchmarks/reviewer-contract/ remain in the
source repository and are outside this synthetic dataset's explicit export list.
The source repository also contains
workflow screening experiments
outside this dataset's export list. These source-checkout profiles explore
failure triage, excerpt relevance, and claim support. The
profile evaluator guide
explains how to run them; the
September 28, 2026 trial
records their method, outcomes, and limits.
Provenance and updates
The source repository is licensed under MIT.
Its original license applies
to these project-authored fixtures and reports. Model weights are not included.
export-manifest.json records the source commit and each file's SHA-256.
Future studies add dated files and preserve earlier observations and fixture
versions. Cite the dataset revision, fixture version, and report filename when
using a result.
- Downloads last month
- 60