Datasets:
File size: 4,692 Bytes
1f6e4f1 4ae5294 1f6e4f1 4ae5294 1f6e4f1 4ae5294 1f6e4f1 4ae5294 1f6e4f1 4ae5294 1f6e4f1 4ae5294 1f6e4f1 4ae5294 1f6e4f1 4ae5294 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 | ---
license: other
license_name: mit-and-cc-by-4.0
pretty_name: competitive-code-curated
task_categories:
- text-generation
tags:
- code
- competitive-programming
- reinforcement-learning
size_categories:
- 1K<n<10K
configs:
- config_name: default
data_files:
- split: train
path: problems-train.parquet
- split: eval
path: problems-eval.parquet
- split: qualification
path: problems-qualification.parquet
---
# competitive-code-curated
Competitive-programming problems graded on hidden stdin/stdout tests, curated
for verifiable RL by the `reliquary-competitive-code` environment
(reliquary-environments, `environments/code/reliquary_competitive_code`).
Python 3 only. See `NOTICE` for sources and licences: rows from DeepCoder are
MIT, rows added in v1.1 (open-r1/codeforces, Code-Contests-Plus) are CC-BY-4.0.
## Files
| file | rows |
| --- | --- |
| `problems-train.parquet` | 8,792 (6,899 from v1, 1,893 added in v1.1) |
| `problems-eval.parquet` | 166 |
| `problems-qualification.parquet` | 208 |
| `references.parquet` | 9,166 (one passing Python 3 solution per problem; `""` for a v1.1 problem shipped on upstream tests only) |
| `build_report.json` | drop counts and sha256 of the four files |
Problem columns: `problem_id` (16 hex), `statement`, `tests_json`
(`[[stdin, stdout], ...]`), `time_limit_s` (CPU seconds), `origin_json`
(upstream source and row; v1.1 rows add `reference_kind`, `rating`,
`contest_date` and `upstream_time_limit_s`). Tests are stored in row groups and read lazily.
## SFT / RL share of train
`train` is sorted by `problem_id` (the first 16 hex of a sha256 of the
normalised statement). A problem is for SFT when `int(problem_id, 16) <
0x9999999999999999` (60% of the hash space), otherwise for RL, so each share is
a contiguous range of `problems-train.parquet`:
| use | rows of train | problems |
| --- | --- | --- |
| SFT | `[0, 4181)` | 4,181 |
| RL | `[4181, 6899)` | 2,718 |
| SFT (v1.1) | `[6899, 8792)` | 1,893 |
The rule depends only on the problem's identity: a rebuild keeps every problem
on its side. Problems added in v1.1 are all for SFT and appended after the v1
rows (sorted by `problem_id` among themselves), so every v1 row keeps its
index and content.
## Build (2026-10-09)
From DeepCoder-Preview-Dataset @`177913a7` (taco + primeintellect, 21,294
rows), 32 cores, 1 h 10 min:
| step | count |
| --- | --- |
| loaded | 21,294 |
| dropped: statement allows several answers | 2,180 |
| dropped: tests contradict a statement example | 388 |
| dropped: overlaps LiveCodeBench | 37 |
| problems after dedup | 8,509 |
| dropped: no passing Python 3 reference | 1,150 |
| dropped: reference too slow | 46 |
| dropped: references disagree | 35 |
| dropped: too few tests | 5 |
| **curated** | **7,273** (taco 5,316, primeintellect 1,957) |
Most `no passing reference` drops are primeintellect problems whose upstream
solutions are Python 2 (`print x`, `raw_input`).
## Grading
Reward 1.0 when the program passes every test within `time_limit_s` CPU
seconds, else 0.0. Token-wise comparison (`yes`/`no` case-insensitive,
decimals within 1e-6). Problems with several valid answers are excluded:
there is no checker.
## v1.1 extension (2026-10-15)
1,893 problems appended to train, all SFT. Eval and qualification are
unchanged (same sha256). Sources: open-r1/codeforces `verifiable` @`fbe3f6e9`
(rating 1200-2400, before 2024-08, stdio, no checker, not interactive; Python
references from open-r1/codeforces-submissions @`07e7bf31`), Code-Contests-Plus
1x @`96c85054` (stock checkers our comparator reproduces exactly), and v1
problems rejected only because their reference was Python 2 (converted with
2to3). Deduplicated against the v1 pool and across sources (exact match plus
13-gram containment at 0.3), decontaminated against every LiveCodeBench
version and v1's eval and qualification statements.
| | open-r1 | Code-Contests-Plus | Python 2 |
| --- | --- | --- | --- |
| selected | 963 | 1,314 | 412 |
| dropped: too few tests within 4 MiB | 214 | 17 | - |
| dropped: upstream tests not validated | - | 318 | 81 |
| dropped: references disagree / wrong | 14 | 32 | - |
| dropped: figure in statement, no reference | - | 108 | - |
| dropped: other | - | 10 | 2 |
| **appended** | **735** | **829** | **329** |
`reference_kind`: `python_reference` 213, `py2to3` 674, `upstream_tests_only`
1,006. An `upstream_tests_only` problem has no Python solution that passes;
its tests are the upstream's validated ones and a completion is graded on them
alone. Time limit = 4 x the upstream limit, between 1 and 10 s, never below 4 x
a passing reference's CPU time. Up to 32 tests and 4 MiB of tests per problem.
|