Skip to content
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

Switch from SGD optimizer to AdamW for patch classification tasks #690

Merged
merged 9 commits into from
Oct 18, 2024
Merged

Switch from SGD optimizer to AdamW for patch classification tasks #690

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
59fc5d6
change SDG to AdamW, batchsize to 256 for patch classification
nkaenzig Oct 17, 2024
d41bbca
replaced early stopping criteria
nkaenzig Oct 17, 2024
f209897
increased patience for mhist to 100
nkaenzig Oct 17, 2024
5c48217
changed early stopping criteria
nkaenzig Oct 18, 2024
382d1fc
update patience for mhist to follow protocol
nkaenzig Oct 18, 2024
a627162
updated docs
nkaenzig Oct 18, 2024
fea4b73
update leaderboard & plots
nkaenzig Oct 18, 2024
624e7ef
update sentence in protocol docs
nkaenzig Oct 18, 2024
b898973
Merge branch 'main' into sgd-to-adam
nkaenzig Oct 18, 2024
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
6 changes: 2 additions & 4 deletions configs/core/tests/offline/embeddings.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,11 +26,9 @@ model:
out_features: &NUM_CLASSES 2
criterion: torch.nn.CrossEntropyLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: ${oc.env:LR_VALUE, 0.1}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
6 changes: 2 additions & 4 deletions configs/vision/pathology/offline/classification/bach.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,11 +53,9 @@ model:
out_features: &NUM_CLASSES 4
criterion: torch.nn.CrossEntropyLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.000625}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
2 changes: 1 addition & 1 deletion configs/vision/pathology/offline/classification/camelyon16.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ trainer:
filename: best
save_last: true
save_top_k: 1
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryAccuracy}
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryBalancedAccuracy}
mode: &MONITOR_METRIC_MODE ${oc.env:MONITOR_METRIC_MODE, max}
- class_path: lightning.pytorch.callbacks.EarlyStopping
init_args:
Expand Down
2 changes: 1 addition & 1 deletion configs/vision/pathology/offline/classification/camelyon16_small.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ trainer:
filename: best
save_last: true
save_top_k: 1
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryAccuracy}
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryBalancedAccuracy}
mode: &MONITOR_METRIC_MODE ${oc.env:MONITOR_METRIC_MODE, max}
- class_path: lightning.pytorch.callbacks.EarlyStopping
init_args:
Expand Down
8 changes: 3 additions & 5 deletions configs/vision/pathology/offline/classification/crc.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,11 +53,9 @@ model:
out_features: &NUM_CLASSES 9
criterion: torch.nn.CrossEntropyLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.01}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down Expand Up @@ -104,7 +102,7 @@ data:
split: val
dataloaders:
train:
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 4096}
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 256}
num_workers: &N_DATA_WORKERS ${oc.env:N_DATA_WORKERS, 4}
shuffle: true
val:
Expand Down
10 changes: 4 additions & 6 deletions configs/vision/pathology/offline/classification/mhist.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,12 +18,12 @@ trainer:
filename: best
save_last: true
save_top_k: 1
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryAccuracy}
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryBalancedAccuracy}
mode: &MONITOR_METRIC_MODE ${oc.env:MONITOR_METRIC_MODE, max}
- class_path: lightning.pytorch.callbacks.EarlyStopping
init_args:
min_delta: 0
patience: 51
patience: 70
monitor: *MONITOR_METRIC
mode: *MONITOR_METRIC_MODE
- class_path: eva.callbacks.ClassificationEmbeddingsWriter
Expand Down Expand Up @@ -53,11 +53,9 @@ model:
out_features: 1
criterion: torch.nn.BCEWithLogitsLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.000625}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
10 changes: 4 additions & 6 deletions configs/vision/pathology/offline/classification/patch_camelyon.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ trainer:
filename: best
save_last: true
save_top_k: 1
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryAccuracy}
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryBalancedAccuracy}
mode: &MONITOR_METRIC_MODE ${oc.env:MONITOR_METRIC_MODE, max}
- class_path: lightning.pytorch.callbacks.EarlyStopping
init_args:
Expand Down Expand Up @@ -54,11 +54,9 @@ model:
out_features: 1
criterion: torch.nn.BCEWithLogitsLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.01}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down Expand Up @@ -118,7 +116,7 @@ data:
split: test
dataloaders:
train:
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 4096}
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 256}
num_workers: &N_DATA_WORKERS ${oc.env:N_DATA_WORKERS, 4}
shuffle: true
val:
Expand Down
6 changes: 2 additions & 4 deletions configs/vision/pathology/online/classification/bach.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,11 +45,9 @@ model:
out_features: &NUM_CLASSES 4
criterion: torch.nn.CrossEntropyLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.000625}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
8 changes: 3 additions & 5 deletions configs/vision/pathology/online/classification/crc.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,11 +45,9 @@ model:
out_features: &NUM_CLASSES 9
criterion: torch.nn.CrossEntropyLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.01}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down Expand Up @@ -86,7 +84,7 @@ data:
split: val
dataloaders:
train:
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 4096}
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 256}
num_workers: &N_DATA_WORKERS ${oc.env:N_DATA_WORKERS, 4}
shuffle: true
val:
Expand Down
10 changes: 4 additions & 6 deletions configs/vision/pathology/online/classification/mhist.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,12 +17,12 @@ trainer:
filename: best
save_last: true
save_top_k: 1
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryAccuracy}
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryBalancedAccuracy}
mode: &MONITOR_METRIC_MODE ${oc.env:MONITOR_METRIC_MODE, max}
- class_path: lightning.pytorch.callbacks.EarlyStopping
init_args:
min_delta: 0
patience: 51
patience: 70
monitor: *MONITOR_METRIC
mode: *MONITOR_METRIC_MODE
logger:
Expand All @@ -45,11 +45,9 @@ model:
out_features: 1
criterion: torch.nn.BCEWithLogitsLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.000625}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
10 changes: 4 additions & 6 deletions configs/vision/pathology/online/classification/patch_camelyon.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@ trainer:
filename: best
save_last: true
save_top_k: 1
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryAccuracy}
monitor: &MONITOR_METRIC ${oc.env:MONITOR_METRIC, val/BinaryBalancedAccuracy}
mode: &MONITOR_METRIC_MODE ${oc.env:MONITOR_METRIC_MODE, max}
- class_path: lightning.pytorch.callbacks.EarlyStopping
init_args:
Expand Down Expand Up @@ -45,11 +45,9 @@ model:
out_features: 1
criterion: torch.nn.BCEWithLogitsLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.01}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down Expand Up @@ -91,7 +89,7 @@ data:
split: test
dataloaders:
train:
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 4096}
batch_size: &BATCH_SIZE ${oc.env:BATCH_SIZE, 256}
num_workers: &N_DATA_WORKERS ${oc.env:N_DATA_WORKERS, 4}
shuffle: true
val:
Expand Down
6 changes: 2 additions & 4 deletions configs/vision/tests/offline/panda.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,11 +34,9 @@ model:
output_size: &NUM_CLASSES 6
criterion: torch.nn.CrossEntropyLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: &LR_VALUE ${oc.env:LR_VALUE, 0.00004}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
6 changes: 2 additions & 4 deletions configs/vision/tests/offline/patch_camelyon.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,11 +47,9 @@ model:
out_features: 1
criterion: torch.nn.BCEWithLogitsLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: ${oc.env:LR_VALUE, 0.1}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
6 changes: 2 additions & 4 deletions configs/vision/tests/online/patch_camelyon.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,11 +18,9 @@ model:
out_features: 1
criterion: torch.nn.BCEWithLogitsLoss
optimizer:
class_path: torch.optim.SGD
class_path: torch.optim.AdamW
init_args:
lr: ${oc.env:LR_VALUE, 0.1}
momentum: 0.9
weight_decay: 0.0
lr: ${oc.env:LR_VALUE, 0.0003}
lr_scheduler:
class_path: torch.optim.lr_scheduler.CosineAnnealingLR
init_args:
Expand Down
Loading