Add 13 new procedural datasets across 7 categories

New dataset categories: combinatorics, statistics, optimization, and
formal languages. Extended existing algebra, arithmetic, probability,
logic, and graphs packages with complex_advanced, linear_algebra, limits,
number_theory, conditional_probability, set_operations, and job_scheduling.

Each dataset includes config validation, deterministic seeding, custom
scoring, curriculum support, and comprehensive unit tests (92 new tests).
This commit is contained in:
Ritvik19 2026-04-18 16:42:54 +05:30
parent 49b07130b3
commit 6eb252ae32
36 changed files with 3705 additions and 1 deletions

View file

@ -9,13 +9,17 @@ from . import (
arithmetic,
code,
cognition,
combinatorics,
data,
games,
geometry,
graphs,
induction,
languages,
logic,
optimization,
probability,
statistics,
)
from .factory import create_dataset, get_score_answer_fn, register_dataset
from .scoring import cascade_score, float_match, math_match, string_match, strip_latex
@ -28,13 +32,17 @@ __all__ = [
"arithmetic",
"code",
"cognition",
"combinatorics",
"data",
"games",
"geometry",
"graphs",
"languages",
"logic",
"induction",
"optimization",
"probability",
"statistics",
"create_dataset",
"register_dataset",
"get_score_answer_fn",