← Back to runs

PAAD

Last updated 2026-01-29 14:43 · Cached 2026-02-08 16:56

Features Inference Manifest Tiling Training

Stage Summary Pipeline status at a glance Pipeline

Features
Encoders detected: gigapath_ft (3010 files), h-optimus-0 (3010 files), prov-gigapath (3010 files), uni (3010 files), virchow (3010 files), virchow2 (3010 files) Additional encoders: FoundationModel_ViT-L_p16_b2048 (3010 files)
Inference
265 prediction(s) · 200 positive label(s) · dataset tcga
Manifest
3018 rows, 11 columns
Tiling
Tiling manifest covers 3014 case(s)
Training
30 training run(s)

Case Summary Track cohort coverage by encoder Cohort

Cases Inspect case-level tiles and encoders Cases

Manifest Overview Preview the manifest and target distribution Manifest

3018 rows · 9 columns

Target: KRAS_BINARY Target: KRAS

slide_id	KRAS_binary	DMP_ASSAY_ID	split_1	split_2	split_3	split_4	split_5	KRAS
P-0000142-T01-IM3	1	P-0000142-T01-IM3	train	train	test	train	train	Positive
P-0000314-T01-IM3	0	P-0000314-T01-IM3	test	train	train	train	train	Negative
P-0000783-T01-IM3	1	P-0000783-T01-IM3	train	test	train	train	train	Positive
P-0000825-T04-IM7	1	P-0000825-T04-IM7	test	test	train	train	train	Positive
P-0001304-T01-IM3	1	P-0001304-T01-IM3	test	train	train	train	test	Positive

Cross-Validation Performance Compare experiments across validation splits Validation

External Inference Diagnostics Interactive ROC/PR curves for each external inference experiment Inference