[vision] Add problem_type support (huggingface#15851)

* Add problem_type to missing models

* Fix deit test

Co-authored-by: Niels Rogge <nielsrogge@Nielss-MacBook-Pro.local>

src/transformers/models/beit/modeling_beit.py

@@ -22,7 +22,7 @@
 import torch
 import torch.utils.checkpoint
 from torch import nn
-from torch.nn import CrossEntropyLoss, MSELoss
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
 
 from ...activations import ACT2FN
 from ...file_utils import (
@@ -857,14 +857,26 @@ def forward(
 
         loss = None
         if labels is not None:
-            if self.num_labels == 1:
-                #  We are doing regression
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
                 loss_fct = MSELoss()
-                loss = loss_fct(logits.view(-1), labels.view(-1))
-            else:
+                if self.num_labels == 1:
+                    loss = loss_fct(logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(logits, labels)
+            elif self.config.problem_type == "single_label_classification":
                 loss_fct = CrossEntropyLoss()
                 loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
-
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(logits, labels)
         if not return_dict:
             output = (logits,) + outputs[2:]
             return ((loss,) + output) if loss is not None else output

src/transformers/models/deit/modeling_deit.py

@@ -23,7 +23,7 @@
 import torch
 import torch.utils.checkpoint
 from torch import nn
-from torch.nn import CrossEntropyLoss, MSELoss
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
 
 from ...activations import ACT2FN
 from ...file_utils import (
@@ -726,14 +726,26 @@ def forward(
 
         loss = None
         if labels is not None:
-            if self.num_labels == 1:
-                #  We are doing regression
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
                 loss_fct = MSELoss()
-                loss = loss_fct(logits.view(-1), labels.view(-1))
-            else:
+                if self.num_labels == 1:
+                    loss = loss_fct(logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(logits, labels)
+            elif self.config.problem_type == "single_label_classification":
                 loss_fct = CrossEntropyLoss()
                 loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
-
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(logits, labels)
         if not return_dict:
             output = (logits,) + outputs[2:]
             return ((loss,) + output) if loss is not None else output

src/transformers/models/segformer/modeling_segformer.py

@@ -21,7 +21,7 @@
 import torch
 import torch.utils.checkpoint
 from torch import nn
-from torch.nn import CrossEntropyLoss, MSELoss
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
 
 from ...activations import ACT2FN
 from ...file_utils import (
@@ -590,14 +590,26 @@ def forward(
 
         loss = None
         if labels is not None:
-            if self.num_labels == 1:
-                #  We are doing regression
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
                 loss_fct = MSELoss()
-                loss = loss_fct(logits.view(-1), labels.view(-1))
-            else:
+                if self.num_labels == 1:
+                    loss = loss_fct(logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(logits, labels)
+            elif self.config.problem_type == "single_label_classification":
                 loss_fct = CrossEntropyLoss()
                 loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
-
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(logits, labels)
         if not return_dict:
             output = (logits,) + outputs[1:]
             return ((loss,) + output) if loss is not None else output

src/transformers/models/vit/modeling_vit.py

@@ -21,7 +21,7 @@
 import torch
 import torch.utils.checkpoint
 from torch import nn
-from torch.nn import CrossEntropyLoss, MSELoss
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
 
 from ...activations import ACT2FN
 from ...file_utils import (
@@ -747,14 +747,26 @@ def forward(
 
         loss = None
         if labels is not None:
-            if self.num_labels == 1:
-                #  We are doing regression
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+
+            if self.config.problem_type == "regression":
                 loss_fct = MSELoss()
-                loss = loss_fct(logits.view(-1), labels.view(-1))
-            else:
+                if self.num_labels == 1:
+                    loss = loss_fct(logits.squeeze(), labels.squeeze())
+                else:
+                    loss = loss_fct(logits, labels)
+            elif self.config.problem_type == "single_label_classification":
                 loss_fct = CrossEntropyLoss()
                 loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1))
-
+            elif self.config.problem_type == "multi_label_classification":
+                loss_fct = BCEWithLogitsLoss()
+                loss = loss_fct(logits, labels)
         if not return_dict:
             output = (logits,) + outputs[2:]
             return ((loss,) + output) if loss is not None else output

tests/deit/test_modeling_deit.py

@@ -17,6 +17,7 @@
 
 import inspect
 import unittest
+import warnings
 
 from transformers import DeiTConfig
 from transformers.file_utils import cached_property, is_torch_available, is_vision_available
@@ -32,6 +33,8 @@
     from torch import nn
 
     from transformers import (
+        MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING,
+        MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
         MODEL_MAPPING,
         DeiTForImageClassification,
         DeiTForImageClassificationWithTeacher,
@@ -379,6 +382,57 @@ def test_training_gradient_checkpointing(self):
             loss = model(**inputs).loss
             loss.backward()
 
+    def test_problem_types(self):
+        config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
+
+        problem_types = [
+            {"title": "multi_label_classification", "num_labels": 2, "dtype": torch.float},
+            {"title": "single_label_classification", "num_labels": 1, "dtype": torch.long},
+            {"title": "regression", "num_labels": 1, "dtype": torch.float},
+        ]
+
+        for model_class in self.all_model_classes:
+            if (
+                model_class
+                not in [
+                    *get_values(MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING),
+                    *get_values(MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING),
+                ]
+                or model_class.__name__ == "DeiTForImageClassificationWithTeacher"
+            ):
+                continue
+
+            for problem_type in problem_types:
+                with self.subTest(msg=f"Testing {model_class} with {problem_type['title']}"):
+
+                    config.problem_type = problem_type["title"]
+                    config.num_labels = problem_type["num_labels"]
+
+                    model = model_class(config)
+                    model.to(torch_device)
+                    model.train()
+
+                    inputs = self._prepare_for_class(inputs_dict, model_class, return_labels=True)
+
+                    if problem_type["num_labels"] > 1:
+                        inputs["labels"] = inputs["labels"].unsqueeze(1).repeat(1, problem_type["num_labels"])
+
+                    inputs["labels"] = inputs["labels"].to(problem_type["dtype"])
+
+                    # This tests that we do not trigger the warning form PyTorch "Using a target size that is different
+                    # to the input size. This will likely lead to incorrect results due to broadcasting. Please ensure
+                    # they have the same size." which is a symptom something in wrong for the regression problem.
+                    # See https://github.com/huggingface/transformers/issues/11780
+                    with warnings.catch_warnings(record=True) as warning_list:
+                        loss = model(**inputs).loss
+                    for w in warning_list:
+                        if "Using a target size that is different to the input size" in str(w.message):
+                            raise ValueError(
+                                f"Something is going wrong in the regression problem: intercepted {w.message}"
+                            )
+
+                    loss.backward()
+
     def test_for_image_classification(self):
         config_and_inputs = self.model_tester.prepare_config_and_inputs()
         self.model_tester.create_and_check_for_image_classification(*config_and_inputs)

tests/test_modeling_common.py

@@ -1873,7 +1873,10 @@ def test_problem_types(self):
         ]
 
         for model_class in self.all_model_classes:
-            if model_class not in get_values(MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING):
+            if model_class not in [
+                *get_values(MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING),
+                *get_values(MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING),
+            ]:
                 continue
 
             for problem_type in problem_types: