Synthetic data factory: procedural and deterministic
Collection
Deterministic project-owned synthetic data: oracle simulation records and verified code-repair batches. • 2 items • Updated
source_file stringclasses 1
value | source_line int64 1 132 | record_json stringlengths 8.77k 45.1k |
|---|---|---|
candidates.jsonl | 1 | {"candidate_prediction":{"method":"inverse of the generator's own mutation (original span restored)","predicted_repair":{"files":{"program.py":"def merge(arr: list[int]) -> list[int]:\n \"\"\"Return a sorted array.\n >>> merge([10,9,8,7,6,5,4,3,2,1])\n [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]\n >>> merge([1,2,3,4,5,... |
candidates.jsonl | 2 | {"candidate_prediction":{"method":"inverse of the generator's own mutation (original span restored)","predicted_repair":{"files":{"program.py":"def is_isogram(string: str) -> bool:\n \"\"\"\n An isogram is a word in which no letter is repeated.\n Examples of isograms are uncopyrightable and ambidextrously.\n ... |
candidates.jsonl | 3 | {"candidate_prediction":{"method":"inverse of the generator's own mutation (original span restored)","predicted_repair":{"files":{"program.py":"def count_vowels(s: str) -> int:\n \"\"\"\n Count the number of vowels in a given string.\n\n :param s: Input string to count vowels in.\n :return: Number of vowels... |
candidates.jsonl | 4 | {"candidate_prediction":{"method":"inverse of the generator's own mutation (original span restored)","predicted_repair":{"files":{"program.py":"from __future__ import annotations\n\n\ndef maximum_non_adjacent_sum(nums: list[int]) -> int:\n \"\"\"\n Find the maximum non-adjacent sum of the integers in the nums inp... |
candidates.jsonl | 5 | {"candidate_prediction":{"method":"inverse of the generator's own mutation (original span restored)","predicted_repair":{"files":{"program.py":"def ceil(x: float) -> int:\n \"\"\"\n Return the ceiling of x as an Integral.\n\n :param x: the number\n :return: the smallest integer >= x.\n\n >>> import math\... |
candidates.jsonl | 6 | {"candidate_prediction":{"method":"inverse of the generator's own mutation (original span restored)","predicted_repair":{"files":{"program.py":"def find_min(numbers: list[int]) -> int:\n \"\"\"\n >>> find_min([1, 2, 3, 4, 5])\n 1\n >>> find_min([5, 5, 5, 5, 5])\n 5\n >>> find_min([5, 5, 5, 5])\n 0\... |
candidates.jsonl | 7 | {"candidate_prediction":{"method":"inverse of the generator's own mutation (original span restored)","predicted_repair":{"files":{"program.py":"from __future__ import annotations\n\n\ndef find_median(nums: list[int | float]) -> float:\n \"\"\"\n This is the implementation of the median.\n :param nums: The list... |
candidates.jsonl | 8 | "{\"candidate_prediction\":{\"method\":\"inverse of the generator's own mutation (original span rest(...TRUNCATED) |
candidates.jsonl | 9 | "{\"candidate_prediction\":{\"method\":\"inverse of the generator's own mutation (original span rest(...TRUNCATED) |
candidates.jsonl | 10 | "{\"candidate_prediction\":{\"method\":\"inverse of the generator's own mutation (original span rest(...TRUNCATED) |
End of preview. Expand in Data Studio
Python Function Repair Records (2026-10-06, deterministic)
Rights & intended use: public research snapshot, not training data. Deterministic, project-owned generation over real MIT-licensed upstream programs (TheAlgorithms/Python @
2067ce6dfb3, MIT — seeprovenance.json):intended_use: training_candidate,project_training_policy: allowedper the factory registry — but this raw snapshot is not training-ready (release-status.json:release_stage: raw_uncurated_public,training_ready: false). Machine-readable record:provenance.json. License: Apache-2.0 for generated content; upstream programs remain MIT by their authors. There is norights.json: the repo's rights schema (v0.1.0) only models hosted-frontier provider routes, so procedural rights are declared here and inprovenance.jsoninstead.
Verification: every candidate carries its execution evidence and the full batch passed independent replay (
evidence/REPLAY.json: status passed).evidence/execution-log.jsonlis harness telemetry, not training rows, and is excluded from the viewer projection.
Contents
132 candidates (150 requested; 18 duplicate mutants + 4 siteless programs skipped):
| file | records | sha256 |
|---|---|---|
data/raw/candidates.jsonl |
132 | 743327ef7a49… |
Viewer projection: data/viewer/records.parquet (132 rows, lossless
source_file / source_line / record_json; Data Studio viewer config).
Reproduce
From the factory repo:
python3 pipelines/code_repair_cli.py generate --catalog catalogs/python-repair-v1 --seed 20261006 --count 150 --out <run-dir>
then replay --run <run-dir> --catalog catalogs/python-repair-v1 --out <replay-dir>.
- Downloads last month
- 53