| 2026-07-14T20:27:39-07:00 healing breadth queue started |
| 2026-07-14T20:27:39-07:00 starting glean_math_keep25 healing from outputs/pruned/glean-0125inst-math-keep25 |
| 2026-07-14T20:39:23-07:00 healing breadth queue started |
| 2026-07-14T20:39:23-07:00 starting glean_math_keep25 healing from outputs/pruned/glean-0125inst-math-keep25 |
| 2026-07-14T20:44:27-07:00 healing breadth queue started |
| 2026-07-14T20:44:27-07:00 starting glean_math_keep25 healing from outputs/pruned/glean-0125inst-math-keep25 |
| 2026-07-14T20:45:07-07:00 queue stopped on an error at line 78 |
| 2026-07-14T20:53:00-07:00 healing breadth queue started |
| 2026-07-14T20:53:00-07:00 starting glean_math_keep25 healing from outputs/pruned/glean-0125inst-math-keep25 |
| 2026-07-14T21:55:23-07:00 glean_math_keep25 healing complete |
| 2026-07-14T21:55:24-07:00 evaluating glean_math_keep25 on full chat GSM8K |
| { |
| "correct": 340, |
| "accuracy": 0.2577710386656558, |
| "finished": 766, |
| "finish_rate": 0.5807429871114481, |
| "mean_completion_tokens": 367.81652767247914 |
| } |
| saved item-level results -> outputs/evals/healing_breadth/glean_math_keep25_seed1224.json |
| 2026-07-14T21:59:08-07:00 glean_math_keep25 result: 0.2577710386656558 |
| 2026-07-14T21:59:08-07:00 starting reap_math_keep75 healing from outputs/pruned/reap48-0125inst-math |
| 2026-07-14T23:27:48-07:00 reap_math_keep75 healing complete |
| 2026-07-14T23:27:48-07:00 evaluating reap_math_keep75 on full chat GSM8K |
| { |
| "correct": 902, |
| "accuracy": 0.6838514025777104, |
| "finished": 1313, |
| "finish_rate": 0.9954510993176648, |
| "mean_completion_tokens": 107.30553449583017 |
| } |
| saved item-level results -> outputs/evals/healing_breadth/reap_math_keep75_seed1224.json |
| 2026-07-14T23:29:27-07:00 reap_math_keep75 result: 0.6838514025777104 |
| 2026-07-14T23:29:27-07:00 materializing math-calibrated Instruct uniform keep-50 checkpoint |
| 2026-07-14T23:29:57-07:00 uniform keep-50 checkpoint ready |
| 2026-07-14T23:29:57-07:00 starting uniform_math_keep50 healing from outputs/pruned/uniform_keep50 |
| 2026-07-15T00:43:17-07:00 uniform_math_keep50 healing complete |
| 2026-07-15T00:43:18-07:00 evaluating uniform_math_keep50 on full chat GSM8K |
| { |
| "correct": 673, |
| "accuracy": 0.510235026535254, |
| "finished": 1302, |
| "finish_rate": 0.9871114480667172, |
| "mean_completion_tokens": 115.02501895375285 |
| } |
| saved item-level results -> outputs/evals/healing_breadth/uniform_math_keep50_seed1224.json |
| 2026-07-15T00:45:11-07:00 uniform_math_keep50 result: 0.510235026535254 |
| 2026-07-15T00:45:11-07:00 healing breadth queue complete |
|
|
