Datasets:
silent-failure-control-v1
9,600 original typed rows; 400 complete synthetic families. Original generated content and labels are CC0-1.0. Source code keeps the repository's MIT license.
Prepared and independently audited only. Not used for training or model evaluation. This addition changes neither the frozen release-v2 mixture nor the data used by the published 2B/9B checkpoints.
Ask one is_silent_failure Noul using only the exact response body string. The transport status is outside model input. Current maintenance, business rejection, a sign-in page replacing requested data, missing explicitly required receipts and unmet delivery requirements are contrasted with valid empty results, recovered history, quoted error text, accepted queued jobs and permitted partial results. Judge transport failures never become negative ground truth. Whole provider/contract families retain all counterfactuals; OOD reserves Chinese wording and different JSON/HTML/text layouts. This controlled grammar does not establish correctness for arbitrary APIs.
| Split | Typed rows |
|---|---|
| train | 6,432 |
| calibration | 264 |
| validation | 360 |
| test | 624 |
| ood | 1,920 |
The community interface reference supplies the task shape. The task instances, labels and instructions are independently authored. No original third-party transcript, whole documentation page or model response is bundled.
Use only state, question, kind and options as model inputs. Targets are reference labels; metadata and auxiliary cases may contain privileged information. The Parquet columns and JSON decoding rules match the root dataset card. Decompressing each raw split recovers its original JSONL bytes exactly. Whole-family splits include correlated counterfactuals; row counts must not be described as independently collected real-world cases.
From the source snapshot's root, regenerate in a new directory with:
python3 -m jev.silent_failure_data --output-dir data/silent-failure-control-v1 --groups 400 --ood-groups 80 --seed 42
No trained-model accuracy, useful compaction, natural-video classification, or general API failure-detection performance is claimed by this release.