Results updatespgaibench(1) / updates | Version | pgaibench-core 0.1.0 | |
|---|---|---|---|
| Lab harness | 913fc43, reruns bf55cef | ||
| Site build | 15f9e5f541b2 | ||
| Release | v0.1.0 (2026-11-11) | ||
| Updated | 2026-11-12T00:04:11Z UTC | ||
| Lab cells | 263 of 264 finished | ||
Every refresh of the provisional PGAIBench-Analyzer results.
v0.1.0 tagged on 2026-11-11; the full list of changes is in the changelog.
Each refresh of the provisional results, newest first, with a dated copy of its data (data/provisional.json at that time). A change to the method or the score is named on the first refresh that includes it.
| data as of (UTC) | lab cells finished | lab cells counted | what changed | release | data |
|---|---|---|---|---|---|
| 2026-11-12T00:04:11Z | 263 of 264 | 263 | no change in the counts | after v0.1.0 | json |
| 2026-11-11T19:07:32Z | 263 of 264 | 263 | v0.1.0 tagged | v0.1.0 | json |
| 2026-11-11T17:25:16Z | 263 of 264 | 263 | headline: three runs per task, averaged (plot, safety table, combined score, findings); one configuration excluded by the inclusion rule; held-out counts, runs side by side and a reasoning-effort table added | v0.1.0 | json |
| 2026-11-11T14:55:14Z | 263 of 264 | 263 | dangerous-change list: the statement that changes the table is shown when the trap pattern does not match; full safety table on desktop | v0.1.0 | json |
| 2026-11-11T14:39:39Z | 263 of 264 | 263 | results page and home: the safety plot first, then its table and findings; the combined score as a secondary view; notes and details below | v0.1.0 | json |
| 2026-11-11T14:21:05Z | 263 of 264 | 263 | safety first: dangerous changes written, where they stopped, tasks done, a four-quadrant plot, every dangerous change listed, pair pages; two L14 cells read by three readers | v0.1.0 | json |
| 2026-11-11T13:50:33Z | 263 of 264 | 263 | rating chart drawn as a picture (SVG) with a table under it; disk-reclaim task: a full rewrite with a bounded lock wait is right, with a reservation; text patterns match only SQL that runs | v0.1.0 | json |
| 2026-11-11T12:51:31Z | 263 of 264 | 263 | pending-review cells read by two readers; classes right with a reservation and correct stop added; row count success checks compare like with like; one rerun cell (D1) judged again: correct | v0.1.0 | json |
| 2026-11-11T07:35:10Z | 263 of 264 | 263 | more lab cells finished | v0.1.0 | json |
| 2026-11-11T07:18:02Z | 263 of 264 | 262 | more lab cells finished | v0.1.0 | json |
| 2026-11-11T07:06:22Z | 263 of 264 | 263 | scoring rules amended: text patterns on normalised SQL, equal stop gets equal class; 23 lab cells judged again offline; D21 counts | v0.1.0 | json |
| 2026-11-11T05:37:27Z | 263 of 264 | 263 | recorded cells on the rerun list count with a flag; pending review counts 0 points | v0.1.0 | json |
| 2026-11-11T05:10:38Z | 263 of 264 | 220 | rating: one list with places, two scores, public tasks only, escalation +0.25; the gemini-4-argon B5 cell is not requested, not unscorable | v0.1.0 | json |
| 2026-11-11T04:07:50Z | 260 of 264 | 218 | rating, PGAIBench-Analyzer score (v0), added after the freeze | v0.1.0 | json |
| 2026-11-11T03:40:18Z | 246 of 264 | 208 | outcome profile chart added | v0.1.0 | json |
| 2026-11-11T03:18:24Z | 215 of 252 | 181 | more lab cells finished | v0.1.0 | json |
| 2026-11-11T02:48:15Z | 184 of 252 | 152 | more lab cells finished | v0.1.0 | json |
| 2026-11-11T02:17:00Z | 156 of 264 | 125 | more lab cells finished | v0.1.0 | json |