Datasets:
Tasks:
Text Classification
Formats:
parquet
Sub-tasks:
multi-label-classification
Size:
10K - 100K
License:
| name: JS/TS Code Review Decisions | |
| description: 'Typed code review decisions: a JavaScript or TypeScript snippet as state and 13 typed questions | |
| (10 flaw nouls, a has_flaw noul, a primary_flaw choice, a quality score) answered with probability distributions. | |
| Scored with score_decisions.py.' | |
| tasks: | |
| - id: macro_f1_at_0_5 | |
| config: decisions | |
| split: test | |
| - id: accuracy | |
| config: decisions | |
| split: test | |
| - id: log_loss | |
| config: decisions | |
| split: test | |
| lower_is_better: true | |
| - id: brier | |
| config: decisions | |
| split: test | |
| lower_is_better: true | |
| - id: ece | |
| config: decisions | |
| split: test | |
| lower_is_better: true | |