Upload AIMv2ForImageClassification

config.json

@@ -0,0 +1,27 @@
+{
+  "_name_or_path": "apple/aimv2-large-patch14-native",
+  "architectures": [
+    "AIMv2ForImageClassification"
+  ],
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "apple/aimv2-large-patch14-native--configuration_aimv2.AIMv2Config",
+    "AutoModel": "apple/aimv2-large-patch14-native--modeling_aimv2.AIMv2Model",
+    "AutoModelForImageClassification": "modeling_aimv2.AIMv2ForImageClassification",
+    "FlaxAutoModel": "apple/aimv2-large-patch14-native--modeling_flax_aimv2.FlaxAIMv2Model"
+  },
+  "hidden_size": 1024,
+  "intermediate_size": 2816,
+  "model_type": "aimv2",
+  "num_attention_heads": 8,
+  "num_channels": 3,
+  "num_hidden_layers": 24,
+  "num_queries": 256,
+  "patch_size": 14,
+  "projection_dropout": 0.0,
+  "qkv_bias": false,
+  "rms_norm_eps": 1e-05,
+  "torch_dtype": "float32",
+  "transformers_version": "4.46.3",
+  "use_bias": false
+}

configuration_aimv2.py

@@ -0,0 +1,60 @@
+from typing import Any
+
+from transformers.configuration_utils import PretrainedConfig
+
+__all__ = ["AIMv2Config"]
+
+
+class AIMv2Config(PretrainedConfig):
+    """This is the configuration class to store the configuration of an [`AIMv2Model`].
+    Instantiating a configuration with the defaults will yield a similar configuration
+    to that of the [apple/aimv2-large-patch14-native](https://huggingface.co/apple/aimv2-large-patch14-native)
+    Args:
+        hidden_size: Dimension of the hidden representations.
+        intermediate_size: Dimension of the SwiGLU representations.
+        num_hidden_layers: Number of hidden layers in the Transformer.
+        num_attention_heads: Number of attention heads for each attention layer
+            in the Transformer.
+        num_channels: Number of input channels.
+        num_queries: Number of learnable queries in the head.
+        patch_size: Patch size.
+        rms_norm_eps: Epsilon value used for the RMS normalization layer.
+        attention_dropout: Dropout ratio for attention probabilities.
+        projection_dropout: Dropout ratio for the projection layer after the attention.
+        qkv_bias: Whether to add a bias to the queries, keys and values.
+        use_bias: Whether to add a bias in the feed-forward and projection layers.
+        kwargs: Keyword arguments for the [`PretrainedConfig`].
+    """
+
+    model_type: str = "aimv2"
+
+    def __init__(
+        self,
+        hidden_size: int = 1024,
+        intermediate_size: int = 2816,
+        num_hidden_layers: int = 24,
+        num_attention_heads: int = 8,
+        num_channels: int = 3,
+        num_queries: int = 256,
+        patch_size: int = 14,
+        rms_norm_eps: float = 1e-5,
+        attention_dropout: float = 0.0,
+        projection_dropout: float = 0.0,
+        qkv_bias: bool = False,
+        use_bias: bool = False,
+        **kwargs: Any,
+    ):
+        super().__init__(**kwargs)
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.num_channels = num_channels
+        self.num_queries = num_queries
+        self.patch_size = patch_size
+        self.attention_dropout = attention_dropout
+        self.rms_norm_eps = rms_norm_eps
+
+        self.projection_dropout = projection_dropout
+        self.qkv_bias = qkv_bias
+        self.use_bias = use_bias

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:310c1a3ac285e0284e06f0df0b9e3e69fbdafb4d5724471d27c416b73bf41779
+size 1235770128

modeling_aimv2.py

@@ -0,0 +1,308 @@
+from typing import Optional, Tuple, Union
+
+import torch
+from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
+from .configuration_aimv2 import AIMv2Config
+from torch import nn
+from torch.nn import functional as F
+from transformers.modeling_outputs import (
+    BaseModelOutputWithNoAttention,
+    ImageClassifierOutput,
+)
+from transformers.modeling_utils import PreTrainedModel
+
+__all__ = ["AIMv2Model"]
+
+
+def _get_1d_sincos_pos_embed_from_grid(
+    embed_dim: int, pos: torch.Tensor
+) -> torch.Tensor:
+    omega = torch.arange(embed_dim // 2).float()
+    omega /= embed_dim / 2.0
+    omega = 1.0 / 10000**omega  # (D / 2,)
+    pos = pos.reshape(-1)  # (M,)
+    out = pos[:, None] * omega[None, :]  # (M, D / 2), outer product
+    emb_sin, emb_cos = torch.sin(out), torch.cos(out)  # (M, D / 2)
+    emb = torch.concatenate([emb_sin, emb_cos], dim=1)  # (M, D)
+    return emb
+
+
+def get_sincos_pos_embed(h: int, w: int, embed_dim: int) -> torch.Tensor:
+    assert embed_dim % 2 == 0, embed_dim
+    grid_h = torch.arange(h).float()
+    grid_w = torch.arange(w).float()
+    grid = torch.meshgrid(grid_w, grid_h, indexing="xy")
+    grid = torch.stack(grid, dim=0)
+    grid = grid.reshape([2, 1, h, w])
+    emb_h = _get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[0])
+    emb_w = _get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[1])
+    pos_embed = torch.concatenate([emb_h, emb_w], dim=1)  # (H * W, D)
+    return pos_embed
+
+
+class RMSNorm(nn.Module):
+    def __init__(self, dim: int, eps: float = 1e-6):
+        super().__init__()
+        self.weight = nn.Parameter(torch.ones(dim))
+        self.eps = eps
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        output = self._norm(x.float()).type_as(x)
+        return output * self.weight
+
+    def extra_repr(self) -> str:
+        return f"{tuple(self.weight.shape)}, eps={self.eps}"
+
+    def _norm(self, x: torch.Tensor) -> torch.Tensor:
+        return x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)
+
+
+class AIMv2SwiGLUFFN(nn.Module):
+    def __init__(self, config: AIMv2Config):
+        super().__init__()
+        hidden_features = config.intermediate_size
+        in_features = config.hidden_size
+        bias = config.use_bias
+
+        self.fc1 = nn.Linear(in_features, hidden_features, bias=bias)
+        self.fc2 = nn.Linear(hidden_features, in_features, bias=bias)
+        self.fc3 = nn.Linear(in_features, hidden_features, bias=bias)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = F.silu(self.fc1(x)) * self.fc3(x)
+        x = self.fc2(x)
+        return x
+
+
+class AIMv2PatchEmbed(nn.Module):
+    def __init__(self, config: AIMv2Config):
+        super().__init__()
+        self.proj = nn.Conv2d(
+            config.num_channels,
+            config.hidden_size,
+            kernel_size=(config.patch_size, config.patch_size),
+            stride=(config.patch_size, config.patch_size),
+        )
+        self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = self.proj(x).flatten(2).transpose(1, 2)
+        x = self.norm(x)
+        return x
+
+
+class AIMv2ViTPreprocessor(nn.Module):
+    def __init__(self, config: AIMv2Config):
+        super().__init__()
+        self.patch_h = config.patch_size
+        self.patch_w = config.patch_size
+        self.embed_dim = config.hidden_size
+
+        self.patchifier = AIMv2PatchEmbed(config)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        _, _, H, W = x.shape
+        tokens = self.patchifier(x)
+        pos_embed = get_sincos_pos_embed(
+            H // self.patch_h, W // self.patch_w, embed_dim=self.embed_dim
+        )
+        tokens = tokens + pos_embed
+        return tokens
+
+
+class AIMv2Attention(nn.Module):
+    def __init__(self, config: AIMv2Config):
+        super().__init__()
+        dim = config.hidden_size
+
+        self.num_heads = config.num_attention_heads
+        self.qkv = nn.Linear(dim, dim * 3, bias=config.qkv_bias)
+        self.attn_drop = nn.Dropout(config.attention_dropout)
+        self.proj = nn.Linear(dim, dim, bias=config.use_bias)
+        self.proj_drop = nn.Dropout(config.projection_dropout)
+
+    def forward(
+        self, x: torch.Tensor, mask: Optional[torch.Tensor] = None
+    ) -> torch.Tensor:
+        B, N, C = x.shape
+        qkv = (
+            self.qkv(x)
+            .reshape(B, N, 3, self.num_heads, C // self.num_heads)
+            .permute(2, 0, 3, 1, 4)
+        )
+        q, k, v = qkv.unbind(0)
+
+        x = F.scaled_dot_product_attention(q, k, v, attn_mask=mask)
+        x = x.transpose(1, 2).contiguous().reshape(B, N, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x
+
+
+class AIMv2Block(nn.Module):
+    def __init__(self, config: AIMv2Config):
+        super().__init__()
+        self.attn = AIMv2Attention(config)
+        self.norm_1 = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+        self.mlp = AIMv2SwiGLUFFN(config)
+        self.norm_2 = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(
+        self, x: torch.Tensor, mask: Optional[torch.Tensor] = None
+    ) -> torch.Tensor:
+        x = x + self.attn(self.norm_1(x), mask)
+        x = x + self.mlp(self.norm_2(x))
+        return x
+
+
+class AIMv2Transformer(nn.Module):
+    def __init__(self, config: AIMv2Config):
+        super().__init__()
+        self.blocks = nn.ModuleList(
+            [AIMv2Block(config) for _ in range(config.num_hidden_layers)]
+        )
+        self.post_trunk_norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
+
+    def forward(
+        self,
+        tokens: torch.Tensor,
+        mask: Optional[torch.Tensor] = None,
+        output_hidden_states: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor, ...]]]:
+        hidden_states = () if output_hidden_states else None
+        for block in self.blocks:
+            tokens = block(tokens, mask)
+            if output_hidden_states:
+                hidden_states += (tokens,)
+        tokens = self.post_trunk_norm(tokens)
+        return tokens, hidden_states
+
+
+class AIMv2PretrainedModel(PreTrainedModel):
+    config_class = AIMv2Config
+    base_model_prefix = "aimv2"
+    main_input_name = "pixel_values"
+    _supports_sdpa = True
+
+
+class AIMv2Model(AIMv2PretrainedModel):
+    def __init__(self, config: AIMv2Config):
+        super().__init__(config)
+        self.preprocessor = AIMv2ViTPreprocessor(config)
+        self.trunk = AIMv2Transformer(config)
+
+    def forward(
+        self,
+        pixel_values: torch.Tensor,
+        mask: Optional[torch.Tensor] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[
+        Tuple[torch.Tensor],
+        Tuple[torch.Tensor, Tuple[torch.Tensor, ...]],
+        BaseModelOutputWithNoAttention,
+    ]:
+        if output_hidden_states is None:
+            output_hidden_states = self.config.output_hidden_states
+        if return_dict is None:
+            return_dict = self.config.use_return_dict
+
+        x = self.preprocessor(pixel_values)
+        x, hidden_states = self.trunk(
+            x, mask, output_hidden_states=output_hidden_states
+        )
+
+        if not return_dict:
+            res = (x,)
+            res += (hidden_states,) if output_hidden_states else ()
+            return res
+
+        return BaseModelOutputWithNoAttention(
+            last_hidden_state=x,
+            hidden_states=hidden_states,
+        )
+
+
+class AIMv2ForImageClassification(AIMv2PretrainedModel):
+    def __init__(self, config: AIMv2Config):
+        super().__init__(config)
+
+        self.num_labels = config.num_labels
+        self.aimv2 = AIMv2Model(config)
+
+        # Classifier head
+        self.classifier = (
+            nn.Linear(config.hidden_size, config.num_labels)
+            if config.num_labels > 0
+            else nn.Identity()
+        )
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(
+        self,
+        pixel_values: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[tuple, ImageClassifierOutput]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the image classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
+
+        outputs = self.aimv2(
+            pixel_values,
+            mask=head_mask,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        sequence_output = outputs[0]
+
+        logits = self.classifier(sequence_output[:, 0, :])
+
+        loss = None
+        if labels is not None:
+            # move labels to correct device to enable model parallelism
+            labels = labels.to(logits.device)
+            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()
+                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
+
+        return ImageClassifierOutput(
+            loss=loss,
+            logits=logits,
+            hidden_states=outputs.hidden_states,
+            # attentions=outputs.attentions,
+        )