Datasets5
Drop a YAML file into acute/data/ to register a new dataset. The watcher uploads it to MinIO and queues GRU + LSTM training jobs automatically.
| Name | Noise % | Subseq % | Rows | Experiments | Status | Registered | |
|---|---|---|---|---|---|---|---|
| threat_n10_z25_s25_labeled datasets/threat_n10_z25_s25_labeled.yaml | 25 | 25 | 26 | 2/2 | complete | today | Experiments → |
| threat_n10_z25_s25 datasets/threat_n10_z25_s25.yaml | 25 | 25 | — | 2/2 | complete | today | Experiments → |
| threat_n10_z0_s25 datasets/threat_n10_z0_s25.yaml | 0 | 25 | — | 0/2 | 2 failed | today | Experiments → |
| threat_n10_z0_s25 datasets/threat_n10_z0_s25.yaml | 0 | 25 | — | 0/2 | 2 failed | today | Experiments → |
| threat_n3966_z0_s25 datasets/threat_n3966_z0_s25.yaml | 0 | 25 | — | 2/2 | complete | today | Experiments → |
To register a new dataset: generate with
python soc.py --num N --noise Z --subsequence S --encode --label in the sec repo, then copy the output YAML to acute/data/. The watcher handles the rest.