Enable dynamic resolution input for Swin Transformer and variants (#3…

…0656)

* add interpolation of positional encoding support to swin

* add style changes

* use default image processor and make size a dictionary

Co-authored-by: amyeroberts <[email protected]>

* remove logits testing

Co-authored-by: amyeroberts <[email protected]>

* Refactor image size validation logic when interpolation is disabled

Co-authored-by: amyeroberts <[email protected]>

* remove asserts in modeling

Co-authored-by: amyeroberts <[email protected]>

* add dynamic resolution input support to swinv2

* change size to ensure interpolation encoding path is triggered

* set interpolate_pos_encoding default value to False

Co-authored-by: amyeroberts <[email protected]>

* set interpolate_pos_encoding default value to False

Co-authored-by: amyeroberts <[email protected]>

* set interpolate_pos_encoding default value to False

Co-authored-by: amyeroberts <[email protected]>

* set interpolate_pos_encoding default value to False

Co-authored-by: amyeroberts <[email protected]>

* set interpolate_pos_encoding default value to False

Co-authored-by: amyeroberts <[email protected]>

* set interpolate_pos_encoding default value to False

Co-authored-by: amyeroberts <[email protected]>

* set interpolate_pos_encoding default value to False

Co-authored-by: amyeroberts <[email protected]>

* set interpolate_pos_encoding default value to False

* add dynamic resolution input to donut swin

* add dynamic resolution input to maskformer swin

---------

Co-authored-by: amyeroberts <[email protected]>

src/transformers/models/donut/modeling_donut_swin.py

@@ -166,10 +166,48 @@ def __init__(self, config, use_mask_token=False):
         self.norm = nn.LayerNorm(config.embed_dim)
         self.dropout = nn.Dropout(config.hidden_dropout_prob)
 
+    def interpolate_pos_encoding(self, embeddings: torch.Tensor, height: int, width: int) -> torch.Tensor:
+        """
+        This method allows to interpolate the pre-trained position encodings, to be able to use the model on higher
+        resolution images.
+
+        Source:
+        https://github.com/facebookresearch/dino/blob/de9ee3df6cf39fac952ab558447af1fa1365362a/vision_transformer.py#L174
+        """
+
+        num_patches = embeddings.shape[1] - 1
+        num_positions = self.position_embeddings.shape[1] - 1
+        if num_patches == num_positions and height == width:
+            return self.position_embeddings
+        class_pos_embed = self.position_embeddings[:, 0]
+        patch_pos_embed = self.position_embeddings[:, 1:]
+        dim = embeddings.shape[-1]
+        h0 = height // self.config.patch_size
+        w0 = width // self.config.patch_size
+        # we add a small number to avoid floating point error in the interpolation
+        # see discussion at https://github.com/facebookresearch/dino/issues/8
+        h0, w0 = h0 + 0.1, w0 + 0.1
+        patch_pos_embed = patch_pos_embed.reshape(1, int(math.sqrt(num_positions)), int(math.sqrt(num_positions)), dim)
+        patch_pos_embed = patch_pos_embed.permute(0, 3, 1, 2)
+        patch_pos_embed = nn.functional.interpolate(
+            patch_pos_embed,
+            scale_factor=(h0 / math.sqrt(num_positions), w0 / math.sqrt(num_positions)),
+            mode="bicubic",
+            align_corners=False,
+        )
+        patch_pos_embed = patch_pos_embed.permute(0, 2, 3, 1).view(1, -1, dim)
+        return torch.cat((class_pos_embed.unsqueeze(0), patch_pos_embed), dim=1)
+
     def forward(
-        self, pixel_values: Optional[torch.FloatTensor], bool_masked_pos: Optional[torch.BoolTensor] = None
+        self,
+        pixel_values: Optional[torch.FloatTensor],
+        bool_masked_pos: Optional[torch.BoolTensor] = None,
+        interpolate_pos_encoding: bool = False,
     ) -> Tuple[torch.Tensor]:
-        embeddings, output_dimensions = self.patch_embeddings(pixel_values)
+        _, num_channels, height, width = pixel_values.shape
+        embeddings, output_dimensions = self.patch_embeddings(
+            pixel_values, interpolate_pos_encoding=interpolate_pos_encoding
+        )
         embeddings = self.norm(embeddings)
         batch_size, seq_len, _ = embeddings.size()
 
@@ -180,7 +218,10 @@ def forward(
             embeddings = embeddings * (1.0 - mask) + mask_tokens * mask
 
         if self.position_embeddings is not None:
-            embeddings = embeddings + self.position_embeddings
+            if interpolate_pos_encoding:
+                embeddings = embeddings + self.interpolate_pos_encoding(embeddings, height, width)
+            else:
+                embeddings = embeddings + self.position_embeddings
 
         embeddings = self.dropout(embeddings)
 
@@ -219,14 +260,21 @@ def maybe_pad(self, pixel_values, height, width):
             pixel_values = nn.functional.pad(pixel_values, pad_values)
         return pixel_values
 
-    def forward(self, pixel_values: Optional[torch.FloatTensor]) -> Tuple[torch.Tensor, Tuple[int]]:
+    def forward(
+        self, pixel_values: Optional[torch.FloatTensor], interpolate_pos_encoding: bool = False
+    ) -> Tuple[torch.Tensor, Tuple[int]]:
         _, num_channels, height, width = pixel_values.shape
         if num_channels != self.num_channels:
             raise ValueError(
                 "Make sure that the channel dimension of the pixel values match with the one set in the configuration."
             )
         # pad the input to be divisible by self.patch_size, if needed
         pixel_values = self.maybe_pad(pixel_values, height, width)
+        if not interpolate_pos_encoding and (height != self.image_size[0] or width != self.image_size[1]):
+            raise ValueError(
+                f"Input image size ({height}*{width}) doesn't match model"
+                f" ({self.image_size[0]}*{self.image_size[1]})."
+            )
         embeddings = self.projection(pixel_values)
         _, _, height, width = embeddings.shape
         output_dimensions = (height, width)
@@ -849,6 +897,8 @@ def _init_weights(self, module):
         output_hidden_states (`bool`, *optional*):
             Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
             more detail.
+        interpolate_pos_encoding (`bool`, *optional*, defaults to `False`):
+            Whether to interpolate the pre-trained position encodings.
         return_dict (`bool`, *optional*):
             Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
 """
@@ -899,6 +949,7 @@ def forward(
         head_mask: Optional[torch.FloatTensor] = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
+        interpolate_pos_encoding: bool = False,
         return_dict: Optional[bool] = None,
     ) -> Union[Tuple, DonutSwinModelOutput]:
         r"""
@@ -921,7 +972,9 @@ def forward(
         # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
         head_mask = self.get_head_mask(head_mask, len(self.config.depths))
 
-        embedding_output, input_dimensions = self.embeddings(pixel_values, bool_masked_pos=bool_masked_pos)
+        embedding_output, input_dimensions = self.embeddings(
+            pixel_values, bool_masked_pos=bool_masked_pos, interpolate_pos_encoding=interpolate_pos_encoding
+        )
 
         encoder_outputs = self.encoder(
             embedding_output,

src/transformers/models/maskformer/modeling_maskformer_swin.py

@@ -163,12 +163,50 @@ def __init__(self, config):
         self.norm = nn.LayerNorm(config.embed_dim)
         self.dropout = nn.Dropout(config.hidden_dropout_prob)
 
-    def forward(self, pixel_values):
-        embeddings, output_dimensions = self.patch_embeddings(pixel_values)
+    def interpolate_pos_encoding(self, embeddings: torch.Tensor, height: int, width: int) -> torch.Tensor:
+        """
+        This method allows to interpolate the pre-trained position encodings, to be able to use the model on higher
+        resolution images.
+
+        Source:
+        https://github.com/facebookresearch/dino/blob/de9ee3df6cf39fac952ab558447af1fa1365362a/vision_transformer.py#L174
+        """
+
+        num_patches = embeddings.shape[1] - 1
+        num_positions = self.position_embeddings.shape[1] - 1
+        if num_patches == num_positions and height == width:
+            return self.position_embeddings
+        class_pos_embed = self.position_embeddings[:, 0]
+        patch_pos_embed = self.position_embeddings[:, 1:]
+        dim = embeddings.shape[-1]
+        h0 = height // self.config.patch_size
+        w0 = width // self.config.patch_size
+        # we add a small number to avoid floating point error in the interpolation
+        # see discussion at https://github.com/facebookresearch/dino/issues/8
+        h0, w0 = h0 + 0.1, w0 + 0.1
+        patch_pos_embed = patch_pos_embed.reshape(1, int(math.sqrt(num_positions)), int(math.sqrt(num_positions)), dim)
+        patch_pos_embed = patch_pos_embed.permute(0, 3, 1, 2)
+        patch_pos_embed = nn.functional.interpolate(
+            patch_pos_embed,
+            scale_factor=(h0 / math.sqrt(num_positions), w0 / math.sqrt(num_positions)),
+            mode="bicubic",
+            align_corners=False,
+        )
+        patch_pos_embed = patch_pos_embed.permute(0, 2, 3, 1).view(1, -1, dim)
+        return torch.cat((class_pos_embed.unsqueeze(0), patch_pos_embed), dim=1)
+
+    def forward(self, pixel_values, interpolate_pos_encoding):
+        _, num_channels, height, width = pixel_values.shape
+        embeddings, output_dimensions = self.patch_embeddings(
+            pixel_values, interpolate_pos_encoding=interpolate_pos_encoding
+        )
         embeddings = self.norm(embeddings)
 
         if self.position_embeddings is not None:
-            embeddings = embeddings + self.position_embeddings
+            if interpolate_pos_encoding:
+                embeddings = embeddings + self.interpolate_pos_encoding(embeddings, height, width)
+            else:
+                embeddings = embeddings + self.position_embeddings
 
         embeddings = self.dropout(embeddings)
 
@@ -207,14 +245,21 @@ def maybe_pad(self, pixel_values, height, width):
             pixel_values = nn.functional.pad(pixel_values, pad_values)
         return pixel_values
 
-    def forward(self, pixel_values: Optional[torch.FloatTensor]) -> Tuple[torch.Tensor, Tuple[int]]:
+    def forward(
+        self, pixel_values: Optional[torch.FloatTensor], interpolate_pos_encoding: bool = False
+    ) -> Tuple[torch.Tensor, Tuple[int]]:
         _, num_channels, height, width = pixel_values.shape
         if num_channels != self.num_channels:
             raise ValueError(
                 "Make sure that the channel dimension of the pixel values match with the one set in the configuration."
             )
         # pad the input to be divisible by self.patch_size, if needed
         pixel_values = self.maybe_pad(pixel_values, height, width)
+        if not interpolate_pos_encoding and (height != self.image_size[0] or width != self.image_size[1]):
+            raise ValueError(
+                f"Input image size ({height}*{width}) doesn't match model"
+                f" ({self.image_size[0]}*{self.image_size[1]})."
+            )
         embeddings = self.projection(pixel_values)
         _, _, height, width = embeddings.shape
         output_dimensions = (height, width)
@@ -780,6 +825,7 @@ def forward(
         head_mask=None,
         output_attentions=None,
         output_hidden_states=None,
+        interpolate_pos_encoding=False,
         return_dict=None,
     ):
         output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
@@ -798,7 +844,9 @@ def forward(
         # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
         head_mask = self.get_head_mask(head_mask, len(self.config.depths))
 
-        embedding_output, input_dimensions = self.embeddings(pixel_values)
+        embedding_output, input_dimensions = self.embeddings(
+            pixel_values, interpolate_pos_encoding=interpolate_pos_encoding
+        )
 
         encoder_outputs = self.encoder(
             embedding_output,

src/transformers/models/swin/modeling_swin.py

@@ -252,10 +252,48 @@ def __init__(self, config, use_mask_token=False):
         self.norm = nn.LayerNorm(config.embed_dim)
         self.dropout = nn.Dropout(config.hidden_dropout_prob)
 
+    def interpolate_pos_encoding(self, embeddings: torch.Tensor, height: int, width: int) -> torch.Tensor:
+        """
+        This method allows to interpolate the pre-trained position encodings, to be able to use the model on higher
+        resolution images.
+
+        Source:
+        https://github.com/facebookresearch/dino/blob/de9ee3df6cf39fac952ab558447af1fa1365362a/vision_transformer.py#L174
+        """
+
+        num_patches = embeddings.shape[1] - 1
+        num_positions = self.position_embeddings.shape[1] - 1
+        if num_patches == num_positions and height == width:
+            return self.position_embeddings
+        class_pos_embed = self.position_embeddings[:, 0]
+        patch_pos_embed = self.position_embeddings[:, 1:]
+        dim = embeddings.shape[-1]
+        h0 = height // self.config.patch_size
+        w0 = width // self.config.patch_size
+        # we add a small number to avoid floating point error in the interpolation
+        # see discussion at https://github.com/facebookresearch/dino/issues/8
+        h0, w0 = h0 + 0.1, w0 + 0.1
+        patch_pos_embed = patch_pos_embed.reshape(1, int(math.sqrt(num_positions)), int(math.sqrt(num_positions)), dim)
+        patch_pos_embed = patch_pos_embed.permute(0, 3, 1, 2)
+        patch_pos_embed = nn.functional.interpolate(
+            patch_pos_embed,
+            scale_factor=(h0 / math.sqrt(num_positions), w0 / math.sqrt(num_positions)),
+            mode="bicubic",
+            align_corners=False,
+        )
+        patch_pos_embed = patch_pos_embed.permute(0, 2, 3, 1).view(1, -1, dim)
+        return torch.cat((class_pos_embed.unsqueeze(0), patch_pos_embed), dim=1)
+
     def forward(
-        self, pixel_values: Optional[torch.FloatTensor], bool_masked_pos: Optional[torch.BoolTensor] = None
+        self,
+        pixel_values: Optional[torch.FloatTensor],
+        bool_masked_pos: Optional[torch.BoolTensor] = None,
+        interpolate_pos_encoding: bool = False,
     ) -> Tuple[torch.Tensor]:
-        embeddings, output_dimensions = self.patch_embeddings(pixel_values)
+        _, num_channels, height, width = pixel_values.shape
+        embeddings, output_dimensions = self.patch_embeddings(
+            pixel_values, interpolate_pos_encoding=interpolate_pos_encoding
+        )
         embeddings = self.norm(embeddings)
         batch_size, seq_len, _ = embeddings.size()
 
@@ -266,7 +304,10 @@ def forward(
             embeddings = embeddings * (1.0 - mask) + mask_tokens * mask
 
         if self.position_embeddings is not None:
-            embeddings = embeddings + self.position_embeddings
+            if interpolate_pos_encoding:
+                embeddings = embeddings + self.interpolate_pos_encoding(embeddings, height, width)
+            else:
+                embeddings = embeddings + self.position_embeddings
 
         embeddings = self.dropout(embeddings)
 
@@ -304,14 +345,21 @@ def maybe_pad(self, pixel_values, height, width):
             pixel_values = nn.functional.pad(pixel_values, pad_values)
         return pixel_values
 
-    def forward(self, pixel_values: Optional[torch.FloatTensor]) -> Tuple[torch.Tensor, Tuple[int]]:
+    def forward(
+        self, pixel_values: Optional[torch.FloatTensor], interpolate_pos_encoding: bool = False
+    ) -> Tuple[torch.Tensor, Tuple[int]]:
         _, num_channels, height, width = pixel_values.shape
         if num_channels != self.num_channels:
             raise ValueError(
                 "Make sure that the channel dimension of the pixel values match with the one set in the configuration."
             )
         # pad the input to be divisible by self.patch_size, if needed
         pixel_values = self.maybe_pad(pixel_values, height, width)
+        if not interpolate_pos_encoding and (height != self.image_size[0] or width != self.image_size[1]):
+            raise ValueError(
+                f"Input image size ({height}*{width}) doesn't match model"
+                f" ({self.image_size[0]}*{self.image_size[1]})."
+            )
         embeddings = self.projection(pixel_values)
         _, _, height, width = embeddings.shape
         output_dimensions = (height, width)
@@ -924,6 +972,8 @@ def _init_weights(self, module):
         output_hidden_states (`bool`, *optional*):
             Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
             more detail.
+        interpolate_pos_encoding (`bool`, *optional*, defaults to `False`):
+            Whether to interpolate the pre-trained position encodings.
         return_dict (`bool`, *optional*):
             Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
 """
@@ -981,6 +1031,7 @@ def forward(
         head_mask: Optional[torch.FloatTensor] = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
+        interpolate_pos_encoding: bool = False,
         return_dict: Optional[bool] = None,
     ) -> Union[Tuple, SwinModelOutput]:
         r"""
@@ -1003,7 +1054,9 @@ def forward(
         # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
         head_mask = self.get_head_mask(head_mask, len(self.config.depths))
 
-        embedding_output, input_dimensions = self.embeddings(pixel_values, bool_masked_pos=bool_masked_pos)
+        embedding_output, input_dimensions = self.embeddings(
+            pixel_values, bool_masked_pos=bool_masked_pos, interpolate_pos_encoding=interpolate_pos_encoding
+        )
 
         encoder_outputs = self.encoder(
             embedding_output,
@@ -1074,6 +1127,7 @@ def forward(
         head_mask: Optional[torch.FloatTensor] = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
+        interpolate_pos_encoding: bool = False,
         return_dict: Optional[bool] = None,
     ) -> Union[Tuple, SwinMaskedImageModelingOutput]:
         r"""
@@ -1113,6 +1167,7 @@ def forward(
             head_mask=head_mask,
             output_attentions=output_attentions,
             output_hidden_states=output_hidden_states,
+            interpolate_pos_encoding=interpolate_pos_encoding,
             return_dict=return_dict,
         )
 
@@ -1156,6 +1211,14 @@ def forward(
     """
     Swin Model transformer with an image classification head on top (a linear layer on top of the final hidden state of
     the [CLS] token) e.g. for ImageNet.
+
+    <Tip>
+
+        Note that it's possible to fine-tune Swin on higher resolution images than the ones it has been trained on, by
+        setting `interpolate_pos_encoding` to `True` in the forward of the model. This will interpolate the pre-trained
+        position embeddings to the higher resolution.
+
+    </Tip>
     """,
     SWIN_START_DOCSTRING,
 )
@@ -1188,6 +1251,7 @@ def forward(
         labels: Optional[torch.LongTensor] = None,
         output_attentions: Optional[bool] = None,
         output_hidden_states: Optional[bool] = None,
+        interpolate_pos_encoding: bool = False,
         return_dict: Optional[bool] = None,
     ) -> Union[Tuple, SwinImageClassifierOutput]:
         r"""
@@ -1203,6 +1267,7 @@ def forward(
             head_mask=head_mask,
             output_attentions=output_attentions,
             output_hidden_states=output_hidden_states,
+            interpolate_pos_encoding=interpolate_pos_encoding,
             return_dict=return_dict,
         )