Datasets:

stan4u
/

formally-verified-c-core-v1

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
copyright_notice: string
data_schema_version: int64
dataset_id: string
dataset_version: string
derivation_family: string
difficulty_bucket: string
features: list<item: string>
  child 0, item: string
has_spectests: bool
license_spdx: string
mode: string
negative_cases: list<item: struct<candidate_source: string, expected: string, name: string>>
  child 0, item: struct<candidate_source: string, expected: string, name: string>
      child 0, candidate_source: string
      child 1, expected: string
      child 2, name: string
non_vacuous: bool
problem: string
provenance: string
reference_solution: string
review_status: string
semantic_family: string
skeleton_c: string
source_content_sha256: string
source_kind: string
source_path: string
source_repository_url: string
source_revision: string
spectests: null
stable_id: string
transformation_description: string
vc_estimate: int64
status: string
splits: struct<test: list<item: string>, train: list<item: string>, validation: list<item: string>>
  child 0, test: list<item: string>
      child 0, item: string
  child 1, train: list<item: string>
      child 0, item: string
  child 2, validation: list<item: string>
      child 0, item: string
schema_version: int64
task_file_sha256: string
counts: struct<test: int64, train: int64, validation: int64>
  child 0, test: int64
  child 1, train: int64
  child 2, validation: int64
split_policy: string
to
{'counts': {'test': Value('int64'), 'train': Value('int64'), 'validation': Value('int64')}, 'dataset_id': Value('string'), 'dataset_version': Value('string'), 'schema_version': Value('int64'), 'split_policy': Value('string'), 'splits': {'test': List(Value('string')), 'train': List(Value('string')), 'validation': List(Value('string'))}, 'status': Value('string'), 'task_file_sha256': 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 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, 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 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, 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
              copyright_notice: string
              data_schema_version: int64
              dataset_id: string
              dataset_version: string
              derivation_family: string
              difficulty_bucket: string
              features: list<item: string>
                child 0, item: string
              has_spectests: bool
              license_spdx: string
              mode: string
              negative_cases: list<item: struct<candidate_source: string, expected: string, name: string>>
                child 0, item: struct<candidate_source: string, expected: string, name: string>
                    child 0, candidate_source: string
                    child 1, expected: string
                    child 2, name: string
              non_vacuous: bool
              problem: string
              provenance: string
              reference_solution: string
              review_status: string
              semantic_family: string
              skeleton_c: string
              source_content_sha256: string
              source_kind: string
              source_path: string
              source_repository_url: string
              source_revision: string
              spectests: null
              stable_id: string
              transformation_description: string
              vc_estimate: int64
              status: string
              splits: struct<test: list<item: string>, train: list<item: string>, validation: list<item: string>>
                child 0, test: list<item: string>
                    child 0, item: string
                child 1, train: list<item: string>
                    child 0, item: string
                child 2, validation: list<item: string>
                    child 0, item: string
              schema_version: int64
              task_file_sha256: string
              counts: struct<test: int64, train: int64, validation: int64>
                child 0, test: int64
                child 1, train: int64
                child 2, validation: int64
              split_policy: string
              to
              {'counts': {'test': Value('int64'), 'train': Value('int64'), 'validation': Value('int64')}, 'dataset_id': Value('string'), 'dataset_version': Value('string'), 'schema_version': Value('int64'), 'split_policy': Value('string'), 'splits': {'test': List(Value('string')), 'train': List(Value('string')), 'validation': List(Value('string'))}, 'status': Value('string'), 'task_file_sha256': Value('string')}
              because column names don't match

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.

Formally Verified C Core-v1

Core-v1 is a project-authored set of 64 fixed-contract C/ACSL function completion tasks for reinforcement-learning environment development and model evaluation. A model receives a complete C translation unit whose target body is replaced by a TODO. The unchanged ACSL contract and surrounding source define the problem; Frama-C WP+RTE supplies the executable reward signal.

33 training tasks
15 validation tasks
16 held-out test tasks
one reference implementation and at least one plausible wrong implementation per task
record-level source hashes, license, origin, transformation, semantic family, derivation family, and review state

Semantic and derivation families are isolated to one split. This prevents a template variant from appearing in both training and evaluation.

Machine-verification evidence

Under Frama-C 33.0 (Arsenic), Why3 1.8.2, Alt-Ergo 2.6.3, and Z3 4.8.12:

64/64 reference implementations proved;
296/296 proof obligations discharged;
84/84 runtime-error obligations discharged;
zero reference solver timeouts; and
64/64 deliberately wrong implementations rejected by the deterministic Frama-C WP+RTE/Qed negative pass, with zero timeouts.

The repository includes the per-task JSONL evidence and a compact release manifest. These are machine-verification results, not an independent human review and not evidence that any RL algorithm improves a model.

Schema

The primary file is tasks.jsonl. Important fields include stable_id, skeleton_c, reference_solution, negative_cases, semantic_family, derivation_family, source_repository_url, source_revision, source_content_sha256, license_spdx, and review_status.

manifest.json defines the frozen family-isolated splits and records the task file checksum.

License and provenance

Core-v1 was authored for this project and is licensed under Apache-2.0. Its deterministic generator is environments/acsl-c/scripts/build_core_v1.py in the canonical repository.

Core-v1 contains no CASP-derived C, skeleton, or reference implementation. The larger CASP corpus used during local environment engineering remains an explicit, non-bundled research adapter because the snapshot available to this project does not retain sufficient per-file origin and license metadata for public redistribution.

Intended use and limitations

Core-v1 is a compact environment seed and release test pack. It is not a comprehensive C verification benchmark, does not represent production C code, and should not be used to claim state-of-the-art model performance. The trusted computing base includes Frama-C, Why3, the selected SMT solvers, the judge runner, and the sandbox runtime.

Canonical repository: https://github.com/stanleyngugi/formally-verified-code-rl

Downloads last month: 32

Total file size:

119 kB

Formally Verified C Core-v1

Contents

Machine-verification evidence

Schema

License and provenance

Intended use and limitations