Spaces:
Runtime error
Runtime error
File size: 3,077 Bytes
3b96cb1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 |
# Copyright (c) OpenMMLab. All rights reserved.
from typing import List, Optional, Tuple, Union
import torch
import torch.nn as nn
from mmengine.dist import all_reduce, get_world_size
from mmengine.model import BaseModule
from mmpretrain.registry import MODELS
@MODELS.register_module()
class LatentPredictHead(BaseModule):
"""Head for latent feature prediction.
This head builds a predictor, which can be any registered neck component.
For example, BYOL and SimSiam call this head and build NonLinearNeck.
It also implements similarity loss between two forward features.
Args:
loss (dict): Config dict for the loss.
predictor (dict): Config dict for the predictor.
init_cfg (dict or List[dict], optional): Initialization config dict.
Defaults to None.
"""
def __init__(self,
loss: dict,
predictor: dict,
init_cfg: Optional[Union[dict, List[dict]]] = None) -> None:
super().__init__(init_cfg=init_cfg)
self.loss_module = MODELS.build(loss)
self.predictor = MODELS.build(predictor)
def loss(self, input: torch.Tensor,
target: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
"""Forward head.
Args:
input (torch.Tensor): NxC input features.
target (torch.Tensor): NxC target features.
Returns:
torch.Tensor: The latent predict loss.
"""
pred = self.predictor([input])[0]
target = target.detach()
loss = self.loss_module(pred, target)
return loss
@MODELS.register_module()
class LatentCrossCorrelationHead(BaseModule):
"""Head for latent feature cross correlation.
Part of the code is borrowed from `script
<https://github.com/facebookresearch/barlowtwins/blob/main/main.py>`_.
Args:
in_channels (int): Number of input channels.
loss (dict): Config dict for module of loss functions.
init_cfg (dict or List[dict], optional): Initialization config dict.
Defaults to None.
"""
def __init__(self,
in_channels: int,
loss: dict,
init_cfg: Optional[Union[dict, List[dict]]] = None) -> None:
super().__init__(init_cfg=init_cfg)
self.world_size = get_world_size()
self.bn = nn.BatchNorm1d(in_channels, affine=False)
self.loss_module = MODELS.build(loss)
def loss(self, input: torch.Tensor, target: torch.Tensor) -> torch.Tensor:
"""Forward head.
Args:
input (torch.Tensor): NxC input features.
target (torch.Tensor): NxC target features.
Returns:
torch.Tensor: The cross correlation loss.
"""
# cross-correlation matrix
cross_correlation_matrix = self.bn(input).T @ self.bn(target)
cross_correlation_matrix.div_(input.size(0) * self.world_size)
all_reduce(cross_correlation_matrix)
loss = self.loss_module(cross_correlation_matrix)
return loss
|