Datasets:
Remove notes below the keep-or-archive rules
Browse files
README.md
CHANGED
|
@@ -67,10 +67,6 @@ A changed task has a new `content_sha256` and is validated again from stage 1.
|
|
| 67 |
| 4 and 5, oracle and NOP | archive if the reward is wrong: the reference solution must get 1, doing nothing must get 0 |
|
| 68 |
| 2, 6, 7 and 8 | never archive; a person judges |
|
| 69 |
|
| 70 |
-
- **Archived tasks are not deleted.** They stay in `archive.parquet`, so a decision can be checked and a fixed task can return.
|
| 71 |
-
- **A failed run is not a failed task.** If a stage could not run for a task, for example because a node failed, the task keeps its previous state.
|
| 72 |
-
- **A result holds for the execution profile it was run on**, which is recorded in the run file: backend, architecture and network mode.
|
| 73 |
-
|
| 74 |
## Updates
|
| 75 |
|
| 76 |
Each validation run is proposed as one pull request. It adds the run file, updates the two Parquets of each affected data source, and states in its description how many tasks were kept and archived, and why.
|
|
|
|
| 67 |
| 4 and 5, oracle and NOP | archive if the reward is wrong: the reference solution must get 1, doing nothing must get 0 |
|
| 68 |
| 2, 6, 7 and 8 | never archive; a person judges |
|
| 69 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
## Updates
|
| 71 |
|
| 72 |
Each validation run is proposed as one pull request. It adds the run file, updates the two Parquets of each affected data source, and states in its description how many tasks were kept and archived, and why.
|