Overview

5 datasets · 142,283 tasks aggregated
Total tasks
142,283
Mean difficulty
5.89
Median difficulty
Avg patch lines
197.4
Avg patch files
4.84

Difficulty distribution All datasets

Label breakdowneasymediumhard
25808 / 68922 / 47551
25,808 68,922 47,551
countminp25medianmeanp75max
142,281 5.89

Score bins

<=314,521 (10.2%)
3.1-525,371 (17.8%)
5.1-754,838 (38.5%)
7.1-839,681 (27.9%)
>87,870 (5.5%)

Languages

python75,943 (53.4%)
go13,659 (9.6%)
javascript11,024 (7.7%)
typescript10,494 (7.4%)
rust8,040 (5.7%)
c6,648 (4.7%)
java5,736 (4.0%)
cpp4,106 (2.9%)
php1,462 (1.0%)
kotlin881 (0.6%)
julia795 (0.6%)
scala460 (0.3%)
elixir411 (0.3%)
swift359 (0.3%)
dart282 (0.2%)

Area / tier

backend66,730 (46.9%)
library49,727 (34.9%)
frontend13,613 (9.6%)
cli12,166 (8.6%)
fullstack41 (0.0%)
framework4 (0.0%)

Top topics

react874 (4.3%)
sympy732 (3.6%)
pandas722 (3.5%)
kubernetes714 (3.5%)
django671 (3.3%)
click602 (3.0%)
xarray490 (2.4%)
sqlglot454 (2.2%)
eslint442 (2.2%)
pydantic434 (2.1%)
swc407 (2.0%)
matplotlib379 (1.9%)
fonttools375 (1.8%)
cobra349 (1.7%)
argparse321 (1.6%)
numpy307 (1.5%)
dask272 (1.3%)
geopandas272 (1.3%)
pennylane243 (1.2%)
qiskit235 (1.2%)
moto232 (1.1%)
clap231 (1.1%)
dvc215 (1.1%)
sphinx213 (1.0%)
globus-sdk213 (1.0%)

Bug classes

incomplete-validation10,626 (20.1%)
missing-feature6,838 (12.9%)
missing-fallback6,411 (12.1%)
type-handling-inconsistency6,202 (11.7%)
missing-metadata-propagation6,138 (11.6%)
wrong-default5,126 (9.7%)
missing-validation1,734 (3.3%)
missing-implementation1,257 (2.4%)
missing-state-propagation1,215 (2.3%)
incomplete-parsing799 (1.5%)
missing-method794 (1.5%)
missing-error-handling791 (1.5%)
race-condition764 (1.4%)
missing-functionality628 (1.2%)
missing-default584 (1.1%)
feature-addition407 (0.8%)
missing-api339 (0.6%)
index-out-of-bounds336 (0.6%)
missing-method-implementation268 (0.5%)
resource-leak209 (0.4%)

Methodology unified difficulty & tagging

Difficulty Scoring (1–10 scale)
Composite weighted score over 5 dimensions:
Patch scope (30%): lines changed, files affected, hunks count
Logic complexity (25%): control-flow depth, branching, algorithmic sophistication
Context breadth (20%): cross-module dependencies, API surface understanding
Test complexity (15%): fixture setup, mock requirements, edge-case coverage
Instruction complexity (10%): problem-statement clarity, implicit requirements
Each dimension is log-scaled, then the weighted sum is mapped to 1–10 and binned into easy (≤4.0), medium (4.1–7.0), hard (>7.0).

Semantic Tagging
Each task is labelled with a 4-tuple [language, area, topic, bug_class]:
language: primary programming language (python, javascript, go, …)
area: architectural tier ∈ {backend, frontend, fullstack, cli, library, framework}
topic: functional domain / library (auth, database, api, numpy, …)
bug_class: root-cause category (logic-error, type-mismatch, race-condition, …)