修改cifar100参数配置

ppcls/configs/ssl/FixMatchCCSSL/FixMatchCCSSL_cifar100_10000_4gpu.yaml

@@ -97,7 +97,6 @@ DataLoader:
         - RandCropImageV2:
             size: [32, 32]
         - NormalizeImage:
-        - Normalize:
             scale: 1.0/255.0
             mean: [0.5071, 0.4867, 0.4408]
             std: [0.2675, 0.2565, 0.2761]
@@ -152,7 +151,7 @@ DataLoader:
             order: hwc
       
       transform_ops_strong2:
-        - RandCropImageV2:
+        - RandomResizedCrop:
             size: [32, 32]
         - RandFlipImage:
             flip_code: 1
@@ -168,8 +167,8 @@ DataLoader:
             p: 0.2
         - NormalizeImage:
             scale: 1.0/255.0
-            mean: [0.5071, 0.4867, 0.4408]
-            std: [0.2675, 0.2565, 0.2761]
+            mean: [0., 0., 0.]
+            std: [1., 1., 1.]
             order: hwc
 
 

ppcls/configs/ssl/FixMatchCCSSL/FixMatchCCSSL_cifar10_4000_4gpu.yaml

@@ -60,7 +60,7 @@ Loss:
   
 UnLabelLoss:
   Train:
-    - CCSSLCeLoss:
+    - CCSSLCELoss:
         weight: 1.
     - SoftSupConLoss:
         weight: 1.0

ppcls/data/dataloader/cifar.py

@@ -89,12 +89,14 @@ class Cifar100(Cifar100_paddle):
                  expand_labels=1,
                  transform_ops=None,
                  transform_ops_weak=None,
-                 transform_ops_strong=None):
+                 transform_ops_strong=None,
+                 transform_ops_strong2=None):
         super().__init__(data_file, mode, None, download, backend)
         assert isinstance(expand_labels, int)
         self._transform_ops = create_operators(transform_ops)
         self._transform_ops_weak = create_operators(transform_ops_weak)
         self._transform_ops_strong = create_operators(transform_ops_strong)
+        self._transform_ops_strong2 = create_operators(transform_ops_strong2)
         self.class_num = 100
 
         labels = []
@@ -117,6 +119,14 @@ class Cifar100(Cifar100_paddle):
             image1 = transform(image, self._transform_ops)
             image1 = image1.transpose((2, 0, 1))
             return (image1, np.int64(label))
+        elif self._transform_ops_weak and self._transform_ops_strong and self._transform_ops_strong2:
+            image2 = transform(image, self._transform_ops_weak)
+            image2 = image2.transpose((2, 0, 1))
+            image3 = transform(image, self._transform_ops_strong)
+            image3 = image3.transpose((2, 0, 1))
+            image4 = transform(image, self._transform_ops_strong2)
+            image4 = image4.transpose((2, 0, 1))
+            return (image2, image3, image4, np.int64(label))
         elif self._transform_ops_weak and self._transform_ops_strong:
             image2 = transform(image, self._transform_ops_weak)
             image2 = image2.transpose((2, 0, 1))

ppcls/engine/train/train_fixmatch_ccssl.py

@@ -41,7 +41,7 @@ def train_epoch_fixmatch_ccssl(engine, epoch_id, print_batch_step):
             engine.unlabel_train_dataloader_iter = iter(engine.unlabel_train_dataloader)
             unlabel_data_batch = engine.unlabel_train_dataloader_iter.next()
 
-        assert len(unlabel_data_batch) == 4
+        assert len(unlabel_data_batch) in [3, 4]
         assert unlabel_data_batch[0].shape == unlabel_data_batch[1].shape == unlabel_data_batch[2].shape
 
         engine.time_info['reader_cost'].update(time.time() - tic)

ppcls/utils/ema.py

@@ -39,4 +39,4 @@ class ExponentialMovingAverage():
         self._update(model, update_fn=lambda e, m: self.decay * e + (1. - self.decay) * m)
 
     def set(self, model):
-        self._update(model, update_fn=lambda e, m: m)
+        self._update(model, update_fn=lambda e, m: m)
\ No newline at end of file