Skip to content

chore: default to stratified_group kfold everywhere #21

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
vojtech-cifka wants to merge 1 commit into
base: master
Choose a base branch
from
Open

chore: default to stratified_group kfold everywhere #21

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions configs/data/dataset.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,8 +14,8 @@ dataset:
test_split_filename: "split_mapping/test_split.csv"
tiling_run_id: "fdf7550a2004474f8c7a05dc0cf1fd86"
filter_tiles_run_id: "9a844082b48d47bfa5d3e7a37d7de268"
stratified_kfold_run_id: "850c81506684450b9af92296acfd045a"
stratified_group_kfold_run_id: "382b41d2fa894514908e8067949c4326"
stratified_kfold_run_id: "e2132cd681e04892b2bfe837a7295b55"
stratified_group_kfold_run_id: "9c2618576b89453893788a7401fef393"
embedding_run_ids:
virchow2: "262d680f52374614baaf19f1e4d05940"
provgigapath: "5099ce57e886411dababfc1234a88b23"
Expand Down
2 changes: 0 additions & 2 deletions configs/experiment/ml/final_linear_provgigapath_adamw.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,8 +7,6 @@ defaults:
embedding_model_name: ProvGigaPath
embedding_dim: 1536
embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.provgigapath}
kfold_strategy: stratified_group
kfold_run_id: ${dataset.mlflow_artifacts.stratified_group_kfold_run_id}
mlflow_artifact_path: linear_classifier_final_provgigapath

# Set after Stage 1 from ProvGigaPath's own AdamW sweep selected by
Expand Down
2 changes: 0 additions & 2 deletions configs/experiment/ml/final_linear_provgigapath_lbfgs.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,8 +7,6 @@ defaults:
embedding_model_name: ProvGigaPath
embedding_dim: 1536
embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.provgigapath}
kfold_strategy: stratified_group
kfold_run_id: ${dataset.mlflow_artifacts.stratified_group_kfold_run_id}
mlflow_artifact_path: linear_classifier_final_provgigapath

# Set after Stage 1 from ProvGigaPath's own LBFGS sweep selected by
Expand Down
3 changes: 0 additions & 3 deletions configs/experiment/ml/train_linear_virchow2_adamw_group_kfold.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,9 +4,6 @@ defaults:
- /ml/task: kfold_linear_classifier
- _self_

kfold_strategy: stratified_group
kfold_run_id: ${dataset.mlflow_artifacts.stratified_group_kfold_run_id}

model:
optimizer: adamw
learning_rate: 1.0e-4
Expand Down
8 changes: 6 additions & 2 deletions configs/ml/task/final_linear_classifier.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,8 +13,12 @@ mode: fit
embedding_model_name: Virchow2
embedding_dim: 2560
embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.virchow2}
kfold_strategy: stratified
kfold_run_id: ${dataset.mlflow_artifacts.stratified_kfold_run_id}
# stratified_group is the default: plain stratified splits tiles from one slide
# across folds, so validation scores are inflated by slide-level leakage. The two
# fields must stay in sync — kfold_strategy only labels the run, kfold_run_id picks
# the data. To use plain stratified folds, override both.
kfold_strategy: stratified_group
kfold_run_id: ${dataset.mlflow_artifacts.stratified_group_kfold_run_id}
filter_tiles_run_id: ${dataset.mlflow_artifacts.filter_tiles_run_id}

train_embedding_uri: runs:/${embedding_run_id}/train/tiles
Expand Down
8 changes: 6 additions & 2 deletions configs/ml/task/kfold_linear_classifier.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,8 +13,12 @@ mode: fit
embedding_model_name: Virchow2
embedding_dim: 2560
embedding_run_id: ${dataset.mlflow_artifacts.embedding_run_ids.virchow2}
kfold_strategy: stratified
kfold_run_id: ${dataset.mlflow_artifacts.stratified_kfold_run_id}
# stratified_group is the default: plain stratified splits tiles from one slide
# across folds, so validation scores are inflated by slide-level leakage. The two
# fields must stay in sync — kfold_strategy only labels the run, kfold_run_id picks
# the data. To use plain stratified folds, override both.
kfold_strategy: stratified_group
kfold_run_id: ${dataset.mlflow_artifacts.stratified_group_kfold_run_id}
filter_tiles_run_id: ${dataset.mlflow_artifacts.filter_tiles_run_id}

train_embedding_uri: runs:/${embedding_run_id}/train/tiles
Expand Down
Loading