src/dinov3/eval/detection/models/transformer_encoder.py

# Copyright (c) Meta Platforms, Inc. and affiliates.
#
# This software may be used and distributed in accordance with
# the terms of the DINOv3 License Agreement.

# https://github.com/facebookresearch/detr/blob/main/models/transformer.py
# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved
"""
DETR Transformer class.

Copy-paste from torch.nn.Transformer with modifications:
    * positional encodings are passed in MHattention
    * extra LN at the end of encoder is removed
    * decoder returns a stack of activations from all decoding layers
"""
from typing import Optional

from torch import Tensor, nn

from ..util.misc import _get_activation_fn, _get_clones


class TransformerEncoder(nn.Module):
    def __init__(self, encoder_layer, num_layers, norm=None):
        super().__init__()
        self.layers = _get_clones(encoder_layer, num_layers)
        self.num_layers = num_layers
        self.norm = norm

    def forward(
        self,
        src,
        mask: Optional[Tensor] = None,
        src_key_padding_mask: Optional[Tensor] = None,
        pos: Optional[Tensor] = None,
    ):
        output = src

        for layer in self.layers:
            output = layer(output, src_mask=mask, src_key_padding_mask=src_key_padding_mask, pos=pos)

        if self.norm is not None:
            output = self.norm(output)

        return output


class TransformerEncoderLayer(nn.Module):
    def __init__(self, d_model, nhead, dim_feedforward=2048, dropout=0.1, activation="relu", normalize_before=False):
        super().__init__()
        # Keeping Dropout 0 in self attention as it makes the eval 10% faster without performance change
        self.self_attn = nn.MultiheadAttention(d_model, nhead, dropout=0, batch_first=True)
        # Implementation of Feedforward model
        self.linear1 = nn.Linear(d_model, dim_feedforward)
        self.dropout = nn.Dropout(dropout)
        self.linear2 = nn.Linear(dim_feedforward, d_model)

        self.norm1 = nn.LayerNorm(d_model)
        self.norm2 = nn.LayerNorm(d_model)
        self.dropout1 = nn.Dropout(dropout)
        self.dropout2 = nn.Dropout(dropout)

        self.activation = _get_activation_fn(activation)
        self.normalize_before = normalize_before

    def with_pos_embed(self, tensor, pos: Optional[Tensor]):
        return tensor if pos is None else tensor + pos

    def forward_post(
        self,
        src,
        src_mask: Optional[Tensor] = None,
        src_key_padding_mask: Optional[Tensor] = None,
        pos: Optional[Tensor] = None,
    ):
        q = k = self.with_pos_embed(src, pos)
        src2 = self.self_attn(
            q, k, value=src, attn_mask=src_mask, key_padding_mask=src_key_padding_mask, need_weights=False
        )[0]
        src = src + self.dropout1(src2)
        src = self.norm1(src)
        src2 = self.linear2(self.dropout(self.activation(self.linear1(src))))
        src = src + self.dropout2(src2)
        src = self.norm2(src)
        return src

    def forward_pre(
        self,
        src,
        src_mask: Optional[Tensor] = None,
        src_key_padding_mask: Optional[Tensor] = None,
        pos: Optional[Tensor] = None,
    ):
        src2 = self.norm1(src)
        q = k = self.with_pos_embed(src2, pos)
        src2 = self.self_attn(
            q, k, value=src2, attn_mask=src_mask, key_padding_mask=src_key_padding_mask, need_weights=False
        )[0]
        src = src + self.dropout1(src2)
        src2 = self.norm2(src)
        src2 = self.linear2(self.dropout(self.activation(self.linear1(src2))))
        src = src + self.dropout2(src2)
        return src

    def forward(
        self,
        src,
        src_mask: Optional[Tensor] = None,
        src_key_padding_mask: Optional[Tensor] = None,
        pos: Optional[Tensor] = None,
    ):
        if self.normalize_before:
            return self.forward_pre(src, src_mask, src_key_padding_mask, pos)
        return self.forward_post(src, src_mask, src_key_padding_mask, pos)