From: Deep active learning for classifying cancer pathology reports
Dataset | Initial training | Validation | Testing | Holdout | Classes | |
---|---|---|---|---|---|---|
Large | 15,000 | 18,032 | 20,036 | 147,284 | 525 (Histology) | 317 (Subsite) |
Small | 1000 | 18,032 | 20,036 | 161,284 | 525 (Histology) | 317 (Subsite) |