Make gradient_checkpointing a training argument

docs/source/model_doc/led.rst

@@ -46,8 +46,8 @@ Tips:
 - LED makes use of *global attention* by means of the ``global_attention_mask`` (see
   :class:`~transformers.LongformerModel`). For summarization, it is advised to put *global attention* only on the first
   ``<s>`` token. For question answering, it is advised to put *global attention* on all tokens of the question.
-- To fine-tune LED on all 16384, it is necessary to enable *gradient checkpointing* by setting
-  ``config.gradient_checkpointing = True``.
+- To fine-tune LED on all 16384, it is necessary to enable *gradient checkpointing* by executing
+  ``model.gradient_checkpointing_enable()``.
 - A notebook showing how to evaluate LED, can be accessed `here
   <https://colab.research.google.com/drive/12INTTR6n64TzS4RrXZxMSXfrOd9Xzamo?usp=sharing>`__.
 - A notebook showing how to fine-tune LED, can be accessed `here

examples/pytorch/language-modeling/README.md

@@ -174,8 +174,3 @@ python run_clm.py --model_type gpt2 --tokenizer_name gpt2 \ --config_overrides="
 ```
 
 This feature is only available in `run_clm.py`, `run_plm.py` and `run_mlm.py`.
-
-This feature can also be used to activate gradient checkpointing by passing:
-```
---config_overrides "gradient_checkpointing=true,use_cache=False"
-```

src/transformers/configuration_utils.py

@@ -19,6 +19,7 @@
 import copy
 import json
 import os
+import warnings
 from typing import Any, Dict, Tuple, Union
 
 from . import __version__
@@ -38,6 +39,9 @@
 logger = logging.get_logger(__name__)
 
 
+NO_SAVE_CONFIG_KEYS = ["_gradient_checkpointing"]
+
+
 class PretrainedConfig(PushToHubMixin):
     r"""
     Base class for all configuration classes. Handles a few parameters common to all models' configurations as well as
@@ -330,6 +334,15 @@ def __init__(self, **kwargs):
         # Drop the transformers version info
         self.transformers_version = kwargs.pop("transformers_version", None)
 
+        # Deal with gradient checkpointing
+        if "gradient_checkpointing" in kwargs:
+            self._gradient_checkpointing = kwargs.pop("gradient_checkpointing")
+            warnings.warn(
+                "Passing `gradient_checkpointing` to a config initialization is deprecated and will be removed in v5 "
+                "Transformers. Using `model.gradient_checkpointing_enable(True) instead, or if you are using the "
+                "`Trainer` API, pass `gradient_checkpointing=True` in your `TrainingArguments`."
+            )
+
         # Additional attributes without default values
         for key, value in kwargs.items():
             try:
@@ -573,6 +586,11 @@ def get_config_dict(
         else:
             logger.info(f"loading configuration file {config_file} from cache at {resolved_config_file}")
 
+        # Backward compatibility: deal with old model files that may have gradient_checkpointing in their config
+        # online
+        if "gradient_checkpointing" in config_dict:
+            config_dict["_gradient_checkpointing"] = config_dict.pop("gradient_checkpointing")
+
         return config_dict, kwargs
 
     @classmethod
@@ -682,6 +700,7 @@ def to_dict(self) -> Dict[str, Any]:
             :obj:`Dict[str, Any]`: Dictionary of all the attributes that make up this configuration instance.
         """
         output = copy.deepcopy(self.__dict__)
+        output = {k: v for k, v in output.items() if k not in NO_SAVE_CONFIG_KEYS}
         if hasattr(self.__class__, "model_type"):
             output["model_type"] = self.__class__.model_type
 

src/transformers/modeling_utils.py

@@ -450,6 +450,7 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, GenerationMixin, PushToHubMix
     _keys_to_ignore_on_save = None
 
     is_parallelizable = False
+    supports_gradient_checkpointing = False
 
     @property
     def dummy_inputs(self) -> Dict[str, torch.Tensor]:
@@ -932,6 +933,18 @@ def prune_heads(self, heads_to_prune: Dict[int, List[int]]):
 
         self.base_model._prune_heads(heads_to_prune)
 
+    def gradient_checkpointing_enable(self, flag: bool = True):
+        """
+        Activates or deactivates gradient checkpointing for the current model.
+
+        Args:
+            flag (:obj:`bool`, `optional`, defaults to :obj:`True`):
+                Will activate gradient checkpointing if :obj:`True`, deactivate it if :obj:`False`.
+        """
+        if not self.supports_gradient_checkpointing and flag:
+            logger.warn(f"{self.__class__.__name__} does not support gradient checkpointing so nothing will happen.")
+        self.config._gradient_checkpointing = flag
+
     def save_pretrained(
         self,
         save_directory: Union[str, os.PathLike],

src/transformers/models/bart/configuration_bart.py

@@ -82,8 +82,6 @@ class BartConfig(PretrainedConfig):
         decoder_layerdrop: (:obj:`float`, `optional`, defaults to 0.0):
             The LayerDrop probability for the decoder. See the `LayerDrop paper <see
             https://arxiv.org/abs/1909.11556>`__ for more details.
-        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            If True, use gradient checkpointing to save memory at the expense of slower backward pass.
         scale_embedding (:obj:`bool`, `optional`, defaults to :obj:`False`):
             Scale embeddings by diving by sqrt(d_model).
         use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
@@ -131,7 +129,6 @@ def __init__(
         init_std=0.02,
         classifier_dropout=0.0,
         scale_embedding=False,
-        gradient_checkpointing=False,
         use_cache=True,
         num_labels=3,
         pad_token_id=1,
@@ -161,7 +158,6 @@ def __init__(
         self.classifier_dropout = classifier_dropout
         self.use_cache = use_cache
         self.num_hidden_layers = encoder_layers
-        self.gradient_checkpointing = gradient_checkpointing
         self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
 
         super().__init__(

src/transformers/models/bart/modeling_bart.py

@@ -471,6 +471,7 @@ def forward(self, hidden_states: torch.Tensor):
 class BartPretrainedModel(PreTrainedModel):
     config_class = BartConfig
     base_model_prefix = "model"
+    supports_gradient_checkpointing = True
     _keys_to_ignore_on_load_unexpected = [r"encoder\.version", r"decoder\.version"]
 
     def _init_weights(self, module):
@@ -782,7 +783,7 @@ def forward(
             if self.training and (dropout_probability < self.layerdrop):  # skip the layer
                 layer_outputs = (None, None)
             else:
-                if getattr(self.config, "gradient_checkpointing", False) and self.training:
+                if getattr(self.config, "_gradient_checkpointing", False) and self.training:
 
                     def create_custom_forward(module):
                         def custom_forward(*inputs):
@@ -1020,12 +1021,11 @@ def forward(
 
             past_key_value = past_key_values[idx] if past_key_values is not None else None
 
-            if getattr(self.config, "gradient_checkpointing", False) and self.training:
+            if getattr(self.config, "_gradient_checkpointing", False) and self.training:
 
                 if use_cache:
                     logger.warning(
-                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
-                        "`use_cache=False`..."
+                        "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/beit/configuration_beit.py

@@ -57,8 +57,6 @@ class BeitConfig(PretrainedConfig):
             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
         layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
             The epsilon used by the layer normalization layers.
-        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            If True, use gradient checkpointing to save memory at the expense of slower backward pass.
         image_size (:obj:`int`, `optional`, defaults to :obj:`224`):
             The size (resolution) of each image.
         patch_size (:obj:`int`, `optional`, defaults to :obj:`16`):

src/transformers/models/beit/modeling_beit.py

@@ -450,7 +450,7 @@ def forward(
 
             layer_head_mask = head_mask[i] if head_mask is not None else None
 
-            if getattr(self.config, "gradient_checkpointing", False) and self.training:
+            if getattr(self.config, "_gradient_checkpointing", False) and self.training:
 
                 def create_custom_forward(module):
                     def custom_forward(*inputs):
@@ -494,6 +494,7 @@ class BeitPreTrainedModel(PreTrainedModel):
 
     config_class = BeitConfig
     base_model_prefix = "beit"
+    supports_gradient_checkpointing = True
 
     def _init_weights(self, module):
         """Initialize the weights"""

src/transformers/models/bert/configuration_bert.py

@@ -92,8 +92,6 @@ class BertConfig(PretrainedConfig):
             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
         layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
             The epsilon used by the layer normalization layers.
-        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            If True, use gradient checkpointing to save memory at the expense of slower backward pass.
         position_embedding_type (:obj:`str`, `optional`, defaults to :obj:`"absolute"`):
             Type of position embedding. Choose one of :obj:`"absolute"`, :obj:`"relative_key"`,
             :obj:`"relative_key_query"`. For positional embeddings use :obj:`"absolute"`. For more information on
@@ -137,7 +135,6 @@ def __init__(
         initializer_range=0.02,
         layer_norm_eps=1e-12,
         pad_token_id=0,
-        gradient_checkpointing=False,
         position_embedding_type="absolute",
         use_cache=True,
         classifier_dropout=None,
@@ -157,7 +154,6 @@ def __init__(
         self.type_vocab_size = type_vocab_size
         self.initializer_range = initializer_range
         self.layer_norm_eps = layer_norm_eps
-        self.gradient_checkpointing = gradient_checkpointing
         self.position_embedding_type = position_embedding_type
         self.use_cache = use_cache
         self.classifier_dropout = classifier_dropout

src/transformers/models/bert/modeling_bert.py

@@ -555,12 +555,11 @@ def forward(
             layer_head_mask = head_mask[i] if head_mask is not None else None
             past_key_value = past_key_values[i] if past_key_values is not None else None
 
-            if getattr(self.config, "gradient_checkpointing", False) and self.training:
+            if getattr(self.config, "_gradient_checkpointing", False) and self.training:
 
                 if use_cache:
                     logger.warning(
-                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
-                        "`use_cache=False`..."
+                        "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False
 
@@ -714,6 +713,7 @@ class BertPreTrainedModel(PreTrainedModel):
     config_class = BertConfig
     load_tf_weights = load_tf_weights_in_bert
     base_model_prefix = "bert"
+    supports_gradient_checkpointing = True
     _keys_to_ignore_on_load_missing = [r"position_ids"]
 
     def _init_weights(self, module):

src/transformers/models/bert_generation/configuration_bert_generation.py

@@ -52,8 +52,6 @@ class BertGenerationConfig(PretrainedConfig):
             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
         layer_norm_eps (:obj:`float`, `optional`, defaults to 1e-12):
             The epsilon used by the layer normalization layers.
-        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            If :obj:`True`, use gradient checkpointing to save memory at the expense of slower backward pass.
         position_embedding_type (:obj:`str`, `optional`, defaults to :obj:`"absolute"`):
             Type of position embedding. Choose one of :obj:`"absolute"`, :obj:`"relative_key"`,
             :obj:`"relative_key_query"`. For positional embeddings use :obj:`"absolute"`. For more information on
@@ -96,7 +94,6 @@ def __init__(
         pad_token_id=0,
         bos_token_id=2,
         eos_token_id=1,
-        gradient_checkpointing=False,
         position_embedding_type="absolute",
         use_cache=True,
         **kwargs
@@ -114,6 +111,5 @@ def __init__(
         self.max_position_embeddings = max_position_embeddings
         self.initializer_range = initializer_range
         self.layer_norm_eps = layer_norm_eps
-        self.gradient_checkpointing = gradient_checkpointing
         self.position_embedding_type = position_embedding_type
         self.use_cache = use_cache

src/transformers/models/big_bird/configuration_big_bird.py

@@ -82,8 +82,6 @@ class BigBirdConfig(PretrainedConfig):
         num_random_blocks (:obj:`int`, `optional`, defaults to 3)
             Each query is going to attend these many number of random blocks. Useful only when :obj:`attention_type ==
             "block_sparse"`.
-        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            If True, use gradient checkpointing to save memory at the expense of slower backward pass.
         classifier_dropout (:obj:`float`, `optional`):
             The dropout ratio for the classification head.
 
@@ -127,7 +125,6 @@ def __init__(
         rescale_embeddings=False,
         block_size=64,
         num_random_blocks=3,
-        gradient_checkpointing=False,
         classifier_dropout=None,
         **kwargs
     ):
@@ -153,7 +150,6 @@ def __init__(
         self.layer_norm_eps = layer_norm_eps
         self.use_cache = use_cache
         self.is_encoder_decoder = is_encoder_decoder
-        self.gradient_checkpointing = gradient_checkpointing
 
         self.rescale_embeddings = rescale_embeddings
         self.attention_type = attention_type

src/transformers/models/big_bird/modeling_big_bird.py

@@ -1598,12 +1598,11 @@ def forward(
             layer_head_mask = head_mask[i] if head_mask is not None else None
             past_key_value = past_key_values[i] if past_key_values is not None else None
 
-            if getattr(self.config, "gradient_checkpointing", False) and self.training:
+            if getattr(self.config, "_gradient_checkpointing", False) and self.training:
 
                 if use_cache:
                     logger.warning(
-                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
-                        "`use_cache=False`..."
+                        "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False
 
@@ -1756,6 +1755,7 @@ class BigBirdPreTrainedModel(PreTrainedModel):
     config_class = BigBirdConfig
     load_tf_weights = load_tf_weights_in_big_bird
     base_model_prefix = "bert"
+    supports_gradient_checkpointing = True
     _keys_to_ignore_on_load_missing = [r"position_ids"]
 
     def _init_weights(self, module):

src/transformers/models/bigbird_pegasus/configuration_bigbird_pegasus.py

@@ -94,8 +94,6 @@ class BigBirdPegasusConfig(PretrainedConfig):
             "block_sparse"`.
         scale_embeddings (:obj:`bool`, `optional`, defaults to :obj:`True`)
             Whether to rescale embeddings with (hidden_size ** 0.5).
-        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            If True, use gradient checkpointing to save memory at the expense of slower backward pass.
 
         Example::
 
@@ -141,7 +139,6 @@ def __init__(
         decoder_start_token_id=2,
         classifier_dropout=0.0,
         scale_embedding=True,
-        gradient_checkpointing=False,
         pad_token_id=0,
         bos_token_id=2,
         eos_token_id=1,
@@ -170,7 +167,6 @@ def __init__(
         self.classifier_dropout = classifier_dropout
         self.use_cache = use_cache
         self.num_hidden_layers = encoder_layers
-        self.gradient_checkpointing = gradient_checkpointing
         self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
 
         # extra config

src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py

@@ -1567,6 +1567,7 @@ def forward(self, hidden_states: torch.Tensor):
 class BigBirdPegasusPreTrainedModel(PreTrainedModel):
     config_class = BigBirdPegasusConfig
     base_model_prefix = "model"
+    supports_gradient_checkpointing = True
 
     def _init_weights(self, module):
         std = self.config.init_std
@@ -1894,7 +1895,7 @@ def forward(
             if self.training and (dropout_probability < self.layerdrop):  # skip the layer
                 layer_outputs = (None, None)
             else:
-                if getattr(self.config, "gradient_checkpointing", False) and self.training:
+                if getattr(self.config, "_gradient_checkpointing", False) and self.training:
 
                     def create_custom_forward(module):
                         def custom_forward(*inputs):
@@ -2225,12 +2226,11 @@ def forward(
 
             past_key_value = past_key_values[idx] if past_key_values is not None else None
 
-            if getattr(self.config, "gradient_checkpointing", False) and self.training:
+            if getattr(self.config, "_gradient_checkpointing", False) and self.training:
 
                 if use_cache:
                     logger.warning(
-                        "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting "
-                        "`use_cache=False`..."
+                        "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                     )
                     use_cache = False
 

src/transformers/models/blenderbot/configuration_blenderbot.py

@@ -78,8 +78,6 @@ class BlenderbotConfig(PretrainedConfig):
         decoder_layerdrop: (:obj:`float`, `optional`, defaults to 0.0):
             The LayerDrop probability for the decoder. See the `LayerDrop paper <see
             https://arxiv.org/abs/1909.11556>`__ for more details.
-        gradient_checkpointing (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            If True, use gradient checkpointing to save memory at the expense of slower backward pass.
         scale_embedding (:obj:`bool`, `optional`, defaults to :obj:`False`):
             Scale embeddings by diving by sqrt(d_model).
         use_cache (:obj:`bool`, `optional`, defaults to :obj:`True`):
@@ -128,7 +126,6 @@ def __init__(
         decoder_start_token_id=1,
         classifier_dropout=0.0,
         scale_embedding=False,
-        gradient_checkpointing=False,
         pad_token_id=0,
         bos_token_id=1,
         eos_token_id=2,
@@ -155,7 +152,6 @@ def __init__(
         self.classifier_dropout = classifier_dropout
         self.use_cache = use_cache
         self.num_hidden_layers = encoder_layers
-        self.gradient_checkpointing = gradient_checkpointing
         self.scale_embedding = scale_embedding  # scale factor will be sqrt(d_model) if True
 
         super().__init__(