| { |
| "name": "P-ARC", |
| "primary_file": "p_arc_dataset.csv", |
| "parquet_file": "p_arc_dataset.parquet", |
| "encoding": "utf-8", |
| "format": "csv", |
| "quoting": "QUOTE_NONNUMERIC", |
| "notes": [ |
| "Cells can exceed Python's default csv.field_size_limit (~128 KiB); call csv.field_size_limit(sys.maxsize) before reading the CSV.", |
| "tasks/<task_id>.json holds each task's train and test pairs in the standard ARC JSON format." |
| ], |
| "columns": [ |
| { |
| "name": "task_id", |
| "type": "string", |
| "description": "Task id, t1 through t50." |
| }, |
| { |
| "name": "task_name", |
| "type": "string", |
| "description": "Short descriptive name of the task, e.g. corner_to_center." |
| }, |
| { |
| "name": "rule", |
| "type": "string", |
| "description": "Natural-language statement of the task's intended rule, reviewed by the authors." |
| }, |
| { |
| "name": "rule_source", |
| "type": "string", |
| "description": "How the rule was written: human_reviewed (redrafted from the task's transform and reviewed), human_annotated (written from scratch), or draft (the original hand-written rule, kept after review)." |
| }, |
| { |
| "name": "train_demonstrations_json", |
| "type": "json_string", |
| "description": "Compact JSON array of training pairs {input, output}; grids are lists of lists of integers 0-9." |
| }, |
| { |
| "name": "test_input_json", |
| "type": "json_string", |
| "description": "Compact JSON grid for the held-out test input." |
| }, |
| { |
| "name": "test_output_json", |
| "type": "json_string", |
| "description": "Compact JSON grid for the held-out test output." |
| }, |
| { |
| "name": "stable_instances_50_json", |
| "type": "json_string", |
| "description": "Compact JSON object {\"train\": [...]} with 50 fixed generator-produced {input, output} pairs." |
| }, |
| { |
| "name": "human_corruptions_json", |
| "type": "json_string", |
| "description": "Compact JSON array of 3 {input, output} pairs with incorrect outputs, made by people: erroneous attempts from the feasibility check or errors designed by the task's creator." |
| }, |
| { |
| "name": "generator_py", |
| "type": "string", |
| "description": "Python source of the task's generator." |
| }, |
| { |
| "name": "verifier_py", |
| "type": "string", |
| "description": "Python source of the task's verifier, which maps an input grid to the correct output grid." |
| } |
| ] |
| } |
|
|
