feat: added LogReg into all self-supervised learning examples (#1325)

* feat: added LogReg into simCLR * fix: fixed typo * Update examples/self_supervised/simCLR.py Co-authored-by: Sergey Kolesnikov <scitator@gmail.com> * feat: added logistic scorer into byol Co-authored-by: Sergey Kolesnikov <scitator@gmail.com>
catalyst-team · Oct 12, 2021 · 93eedf0 · 93eedf0
1 parent 3feea30
commit 93eedf0
Show file tree

Hide file tree

Showing 3 changed files with 36 additions and 2 deletions.
diff --git a/examples/self_supervised/barlow_twins.py b/examples/self_supervised/barlow_twins.py
@@ -70,7 +70,7 @@
         criterion=criterion,
         optimizer=optimizer,
         callbacks=callbacks,
-        loaders=get_loaders(arargs.dataset, args.batch_size, args.num_workersgs),
+        loaders=get_loaders(args.dataset, args.batch_size, args.num_workers),
         verbose=True,
         num_epochs=epochs,
         valid_loader="train",

diff --git a/examples/self_supervised/byol.py b/examples/self_supervised/byol.py
@@ -2,6 +2,7 @@
 import argparse
 
 from common import add_arguments, get_contrastive_model, get_loaders
+from sklearn.linear_model import LogisticRegression
 
 from torch.optim import Adam
 
@@ -50,6 +51,22 @@ def set_requires_grad(model, val):
             ),
             loaders="train",
         ),
+        dl.SklearnModelCallback(
+            feature_key="embedding_origin",
+            target_key="target",
+            train_loader="train",
+            valid_loaders="valid",
+            model_fn=LogisticRegression,
+            predict_key="sklearn_predict",
+            predict_method="predict_proba",
+        ),
+        dl.OptimizerCallback(metric_key="loss"),
+        dl.ControlFlowCallback(
+            dl.AccuracyCallback(
+                target_key="target", input_key="sklearn_predict", topk_args=(1, 3)
+            ),
+            loaders="valid",
+        ),
     ]
 
     runner = SelfSupervisedRunner()

diff --git a/examples/self_supervised/simCLR.py b/examples/self_supervised/simCLR.py
@@ -2,6 +2,7 @@
 import argparse
 
 from common import add_arguments, get_contrastive_model, get_loaders
+from sklearn.linear_model import LogisticRegression
 
 from torch.optim import Adam
 
@@ -26,7 +27,23 @@
     callbacks = [
         dl.CriterionCallback(
             input_key="projection_left", target_key="projection_right", metric_key="loss"
-        )
+        ),
+        dl.SklearnModelCallback(
+            feature_key="embedding_origin",
+            target_key="target",
+            train_loader="train",
+            valid_loaders="valid",
+            model_fn=LogisticRegression,
+            predict_key="sklearn_predict",
+            predict_method="predict_proba",
+        ),
+        dl.OptimizerCallback(metric_key="loss"),
+        dl.ControlFlowCallback(
+            dl.AccuracyCallback(
+                target_key="target", input_key="sklearn_predict", topk_args=(1, 3)
+            ),
+            loaders="valid",
+        ),
     ]
 
     runner = SelfSupervisedRunner()