Dataset Viewer
The dataset viewer is not available for this dataset.
Cannot get the config names for the dataset.
Error code:   ConfigNamesError
Exception:    FileNotFoundError
Message:      Couldn't find any data file at /src/services/worker/ClarusC64/AF-APER-01. Couldn't find 'ClarusC64/AF-APER-01' on the Hugging Face Hub either: FileNotFoundError: Unable to find 'hf://datasets/ClarusC64/AF-APER-01@1c169c1e64f7f45d5108bbdc776b83263536f5e9/data/train.csv' with any supported extension ['.csv', '.tsv', '.json', '.jsonl', '.ndjson', '.parquet', '.geoparquet', '.gpq', '.arrow', '.txt', '.conll', '.conllu', '.tar', '.xml', '.hdf5', '.h5', '.eval', '.toml', '.md', '.lance', '.tsfile', '.vortex', '.fa', '.fasta', '.fna', '.ffn', '.faa', '.frn', '.afa', '.gb', '.gbk', '.genbank', '.fq', '.fastq', '.pdb', '.ent', '.cif', '.mmcif', '.PDB', '.ENT', '.blp', '.bmp', '.dib', '.bufr', '.cur', '.pcx', '.dcx', '.dds', '.ps', '.eps', '.fit', '.fits', '.fli', '.flc', '.ftc', '.ftu', '.gbr', '.gif', '.grib', '.png', '.apng', '.jp2', '.j2k', '.jpc', '.jpf', '.jpx', '.j2c', '.icns', '.ico', '.im', '.iim', '.tif', '.tiff', '.jfif', '.jpe', '.jpg', '.jpeg', '.mpg', '.mpeg', '.msp', '.pcd', '.pxr', '.pbm', '.pgm', '.ppm', '.pnm', '.psd', '.bw', '.rgb', '.rgba', '.sgi', '.ras', '.tga', '.icb', '.vda', '.vst', '.webp', '.wmf', '.emf', '.xbm', '.xpm', '.BLP', '.BMP', '.DIB', '.BUFR', '.CUR', '.PCX', '.DCX', '.DDS', '.PS', '.EPS', '.FIT', '.FITS', '.FLI', '.FLC', '.FTC', '.FTU', '.GBR', '.GIF', '.GRIB', '.PNG', '.APNG', '.JP2', '.J2K', '.JPC', '.JPF', '.JPX', '.J2C', '.ICNS', '.ICO', '.IM', '.IIM', '.TIF', '.TIFF', '.JFIF', '.JPE', '.JPG', '.JPEG', '.MPG', '.MPEG', '.MSP', '.PCD', '.PXR', '.PBM', '.PGM', '.PPM', '.PNM', '.PSD', '.BW', '.RGB', '.RGBA', '.SGI', '.RAS', '.TGA', '.ICB', '.VDA', '.VST', '.WEBP', '.WMF', '.EMF', '.XBM', '.XPM', '.aiff', '.au', '.avr', '.caf', '.flac', '.htk', '.svx', '.mat4', '.mat5', '.mpc2k', '.ogg', '.paf', '.pvf', '.raw', '.rf64', '.sd2', '.sds', '.ircam', '.voc', '.w64', '.wav', '.nist', '.wavex', '.wve', '.xi', '.mp3', '.opus', '.3gp', '.3g2', '.avi', '.asf', '.flv', '.mp4', '.mov', '.m4v', '.mkv', '.webm', '.f4v', '.wmv', '.wma', '.ogm', '.mxf', '.nut', '.AIFF', '.AU', '.AVR', '.CAF', '.FLAC', '.HTK', '.SVX', '.MAT4', '.MAT5', '.MPC2K', '.OGG', '.PAF', '.PVF', '.RAW', '.RF64', '.SD2', '.SDS', '.IRCAM', '.VOC', '.W64', '.WAV', '.NIST', '.WAVEX', '.WVE', '.XI', '.MP3', '.OPUS', '.3GP', '.3G2', '.AVI', '.ASF', '.FLV', '.MP4', '.MOV', '.M4V', '.MKV', '.WEBM', '.F4V', '.WMV', '.WMA', '.OGM', '.MXF', '.NUT', '.glb', '.ply', '.stl', '.GLB', '.PLY', '.STL', '.pdf', '.PDF', '.nii', '.NII', '.zip', '.idx', '.manifest', '.txn']
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/dataset/config_names.py", line 67, in compute_config_names_response
                  config_names = get_dataset_config_names(
                      path=dataset,
                      token=hf_token,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 161, in get_dataset_config_names
                  dataset_module = dataset_module_factory(
                      path,
                  ...<4 lines>...
                      **download_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/load.py", line 1178, in dataset_module_factory
                  raise FileNotFoundError(
                  ...<2 lines>...
                  ) from None
              FileNotFoundError: Couldn't find any data file at /src/services/worker/ClarusC64/AF-APER-01. Couldn't find 'ClarusC64/AF-APER-01' on the Hugging Face Hub either: FileNotFoundError: Unable to find 'hf://datasets/ClarusC64/AF-APER-01@1c169c1e64f7f45d5108bbdc776b83263536f5e9/data/train.csv' with any supported extension ['.csv', '.tsv', '.json', '.jsonl', '.ndjson', '.parquet', '.geoparquet', '.gpq', '.arrow', '.txt', '.conll', '.conllu', '.tar', '.xml', '.hdf5', '.h5', '.eval', '.toml', '.md', '.lance', '.tsfile', '.vortex', '.fa', '.fasta', '.fna', '.ffn', '.faa', '.frn', '.afa', '.gb', '.gbk', '.genbank', '.fq', '.fastq', '.pdb', '.ent', '.cif', '.mmcif', '.PDB', '.ENT', '.blp', '.bmp', '.dib', '.bufr', '.cur', '.pcx', '.dcx', '.dds', '.ps', '.eps', '.fit', '.fits', '.fli', '.flc', '.ftc', '.ftu', '.gbr', '.gif', '.grib', '.png', '.apng', '.jp2', '.j2k', '.jpc', '.jpf', '.jpx', '.j2c', '.icns', '.ico', '.im', '.iim', '.tif', '.tiff', '.jfif', '.jpe', '.jpg', '.jpeg', '.mpg', '.mpeg', '.msp', '.pcd', '.pxr', '.pbm', '.pgm', '.ppm', '.pnm', '.psd', '.bw', '.rgb', '.rgba', '.sgi', '.ras', '.tga', '.icb', '.vda', '.vst', '.webp', '.wmf', '.emf', '.xbm', '.xpm', '.BLP', '.BMP', '.DIB', '.BUFR', '.CUR', '.PCX', '.DCX', '.DDS', '.PS', '.EPS', '.FIT', '.FITS', '.FLI', '.FLC', '.FTC', '.FTU', '.GBR', '.GIF', '.GRIB', '.PNG', '.APNG', '.JP2', '.J2K', '.JPC', '.JPF', '.JPX', '.J2C', '.ICNS', '.ICO', '.IM', '.IIM', '.TIF', '.TIFF', '.JFIF', '.JPE', '.JPG', '.JPEG', '.MPG', '.MPEG', '.MSP', '.PCD', '.PXR', '.PBM', '.PGM', '.PPM', '.PNM', '.PSD', '.BW', '.RGB', '.RGBA', '.SGI', '.RAS', '.TGA', '.ICB', '.VDA', '.VST', '.WEBP', '.WMF', '.EMF', '.XBM', '.XPM', '.aiff', '.au', '.avr', '.caf', '.flac', '.htk', '.svx', '.mat4', '.mat5', '.mpc2k', '.ogg', '.paf', '.pvf', '.raw', '.rf64', '.sd2', '.sds', '.ircam', '.voc', '.w64', '.wav', '.nist', '.wavex', '.wve', '.xi', '.mp3', '.opus', '.3gp', '.3g2', '.avi', '.asf', '.flv', '.mp4', '.mov', '.m4v', '.mkv', '.webm', '.f4v', '.wmv', '.wma', '.ogm', '.mxf', '.nut', '.AIFF', '.AU', '.AVR', '.CAF', '.FLAC', '.HTK', '.SVX', '.MAT4', '.MAT5', '.MPC2K', '.OGG', '.PAF', '.PVF', '.RAW', '.RF64', '.SD2', '.SDS', '.IRCAM', '.VOC', '.W64', '.WAV', '.NIST', '.WAVEX', '.WVE', '.XI', '.MP3', '.OPUS', '.3GP', '.3G2', '.AVI', '.ASF', '.FLV', '.MP4', '.MOV', '.M4V', '.MKV', '.WEBM', '.F4V', '.WMV', '.WMA', '.OGM', '.MXF', '.NUT', '.glb', '.ply', '.stl', '.GLB', '.PLY', '.STL', '.pdf', '.PDF', '.nii', '.NII', '.zip', '.idx', '.manifest', '.txn']

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.

Purpose AF-APER-01 is an open research dataset for testing whether a system can solve relational problems when their representation changes, and whether it can update a solution when constraints change.

Its motivating question is simple: does the model follow the relationships, or a convenient surface pattern?

In the AIRFRAME research programme, Frame-Lock describes persistent reliance on a representation or assumption after it has become misleading. Apertience describes the capacity to recognise, discriminate, release and recondition such commitments. These are motivating concepts, not capabilities proven by this release.

AF-APER-01 v0.3 provides small, deterministic tasks and reference answers. It measures final-answer correctness and selected paired-consistency outcomes. It does not inspect internal model states or establish a causal mechanism of frame release. Dataset contents Split Records Purpose data/train.csv 320 Public development and training examples data/test.csv 80 Public evaluation examples, with answer keys Total 400 Four equally represented challenge families

Each family contains 80 training and 20 test records. Challenge families

  1. Invariant preservation (invariant_preservation) β€” Directed graph problems presented through equivalent node labels or edge-list arrangements. A correct solver preserves the relevant graph relation across representations.
  2. Misleading representation (misleading_representation) β€” Graph problems with surface cues that may distract from the actual directed relationships. The answer must be computed from the graph rather than inferred from labels.
  3. Constraint obstruction (constraint_obstruction) β€” Provider-selection problems in which a revised capacity condition changes the admissible choice. The solver must apply the revised condition and the stated selection rule.
  4. Reconditioning (reconditioning) β€” Route-selection problems in which changed constraints require reconsidering a previously admissible route. The correct route follows the current conditions and specified tie-break rule. These families are intentionally compact. They are not a comprehensive benchmark of agentic planning, general intelligence or long-horizon adaptation. File structure AF-APER-01/ β”œβ”€β”€ data/ β”‚ β”œβ”€β”€ train.csv β”‚ └── test.csv β”œβ”€β”€ scorer.py └── README.md Record format The CSV records contain the challenge text, reference answer, family label, scoring rule and supporting relational metadata. The fields are intended for direct inspection and simple evaluation scripts. Treat expected_answer as a public key, not a secret label.

The required prediction file has two columns: id,prediction AF-APER-01-0001,3 Use the exact IDs in the test CSV. The value above is illustrative, not a verified answer to that ID. Run the scorer Python 3 is sufficient. No third-party packages are required. python scorer.py --predictions predictions.csv --gold data/test.csv The scorer writes a JSON results file and reports overall and per-family accuracy. The v0.3 scorer also reports answer-distribution and baseline diagnostics, counts invalid-format submissions, and includes file hashes for reproducibility.

By default, scored output omits the reference answers. To include them explicitly, use the scorer's --include-answers option. This omission does not conceal the keys: data/test.csv already contains them. Interpreting the scores

  • Accuracy is exact correctness under each record's declared scoring rule.
  • Per-family accuracy helps locate performance differences across the four task types.
  • Majority/constant baselines indicate how much can be achieved without solving the problem. These are not substitutes for stronger rule-based baselines.
  • Paired consistency is meaningful only where paired records are verified equivalent under the stated transformation. Do not interpret reconditioning label-dependent tie-breaks as an invariance test. A high score demonstrates success on these small tasks. It does not establish that a model internally released a frame, learned a carrier-neutral invariant or will generalise to unseen task generators. Recommended evaluation protocol
  1. Load data/test.csv and pass each challenge to the system under evaluation.
  2. Record one final answer per id in predictions.csv.
  3. Run scorer.py against the public gold file.
  4. Report overall accuracy, all four family scores, missing/invalid predictions and any baseline comparisons.
  5. Disclose model version, prompting strategy, tools, sampling settings and whether any records were used for training or prompt selection. For an honest comparison, disclose when the model has access to graph-search code, external solvers or the reference answers. Algorithmic solutions are valid solutions to the tasks, but they do not by themselves support a claim about frame release. Research context AF-APER-01 is an initial dataset asset in the AIRFRAME / Harmotence / Apertience research programme. The longer-term goal is to study relational invariance, consequential differentiation, obstruction and reconditioning under changing admissibility conditions.

This release focuses on observable final-answer behaviour. The broader theoretical claims remain open research questions. Limitations

  • Small synthetic graphs and selection problems can be solved by straightforward algorithms.
  • Repeated generation patterns may allow shortcut learning.
  • The public test split is not blind or contamination-resistant.
  • Scores do not measure hidden reasoning processes or model-internal representation change.
  • Public train/test separation does not guarantee separation of underlying generator templates.
  • The dataset is not a safety certification or a claim of substrate independence. Versioning and citation v0.3 revises challenge construction and scorer handling following audits of answer balance, shortcut baselines, constraint revisions, missing predictions and answer-key disclosure.
Downloads last month
36