MCPcopy Create free account

hub / github.com/Trustworthy-AI-Group/TransferAttack / functions

Functions2,264 in github.com/Trustworthy-AI-Group/TransferAttack

⨍ Functions 2,264◇ Types & classes 551
↓ 96 callersMethodinit_delta
(self, data, **kwargs)
transferattack/attack.py:130
↓ 89 callersMethodget_logits
(self, x, use_noise=False, **kwargs)
transferattack/model_related/faug.py:69
↓ 78 callersMethodupdate_delta
Update adversarial perturbation
transferattack/gradient/gaa.py:146
↓ 72 callersMethodload_state_dict
(self, state_dict)
transferattack/model_related/awt.py:165
↓ 68 callersMethodget_momentum
(self, grad, momentum, **kwargs)
transferattack/gradient/adamsi_fgm.py:39
↓ 61 callersMethodget_grad
(self, loss, delta, **kwargs)
transferattack/model_related/ana.py:115
↓ 60 callersMethodtransform
(self, data, **kwargs)
transferattack/gradient/mig.py:41
↓ 46 callersMethodget_loss
(self, logits, label)
transferattack/gradient/mig.py:45
↓ 38 callersMethodlog
(self, *args, level=INFO)
defense/diffpure/guided_diffusion/logger.py:384
↓ 33 callersMethodload_state_dict
(self, state_dict)
defense/diffpure/score_sde/models/ema.py:103
↓ 33 callersFunctionwrap_model
Add normalization layer with mean and std in training configuration
transferattack/utils.py:37
↓ 33 callersMethodzero_grad
(self)
defense/diffpure/guided_diffusion/fp16_util.py:181
↓ 30 callersMethodapply
(cls, module, name, p, importance_scores=None)
transferattack/ensemble/sasd_ws.py:157
↓ 25 callersFunction_extract_into_tensor
Extract values from a 1-D numpy array for a batch of indices. :param arr: the 1-D numpy array. :param timesteps: a tensor of indices int
defense/diffpure/guided_diffusion/gaussian_diffusion.py:903
↓ 23 callersFunction_cfg
(url='', **kwargs)
transferattack/model_related/setr_networks/tnt.py:19
↓ 23 callersMethodget_loss
Overriden for AGS
transferattack/model_related/ags.py:63
↓ 23 callersMethodrelprop
(self, cam, **kwargs)
transferattack/model_related/ata_vit_utils/Transformer_Explainability/baselines/ViT/ViT_LRP.py:69
↓ 21 callersMethodtransform
(self, data, **kwargs)
transferattack/advanced_objective/cfm.py:89
↓ 21 callersMethodwrite
Write text to stdout (and a file) and optionally flush.
defense/diffpure/utils.py:60
↓ 20 callersMethodbackward
(ctx, grad_LL, grad_LH, grad_HL, grad_HH)
transferattack/model_related/metassa.py:574
↓ 20 callersMethodget_logits
(self, X_adv, model)
transferattack/input_transformation/lpm.py:165
↓ 20 callersFunctionncsn_conv3x3
3x3 convolution with PyTorch initialization. Same as NCSNv1/NCSNv2.
defense/diffpure/score_sde/models/layers.py:108
↓ 20 callersMethodsave
(self)
defense/diffpure/guided_diffusion/train_util.py:240
↓ 19 callersFunctiontrunc_normal_
r"""Fills the input Tensor with values drawn from a truncated normal distribution. The values are effectively drawn from the normal distributi
transferattack/model_related/ata_vit_utils/Transformer_Explainability/baselines/ViT/weight_init.py:42
↓ 18 callersMethod__init__
(self, in_dim, num_units, init_scale=0.1)
defense/diffpure/score_sde/models/layers.py:547
↓ 18 callersMethodget_grad
Overridden for TIM attack.
transferattack/input_transformation/su.py:126
↓ 18 callersMethodupdate_delta
(self, delta, grad, m, v, alpha, **kwargs)
transferattack/input_transformation/idaa.py:141
↓ 17 callersFunctionload_pretrained
(model, cfg=None, num_classes=1000, in_chans=3, filter_fn=None, strict=True)
transferattack/model_related/ata_vit_utils/Transformer_Explainability/baselines/ViT/helpers.py:87
↓ 17 callersFunctionnormalization
Make a standard normalization layer. :param channels: number of input channels. :return: an nn.Module for normalization.
defense/diffpure/guided_diffusion/nn.py:101
↓ 16 callersFunctionclamp
(x, x_min, x_max)
transferattack/utils.py:68
↓ 16 callersFunctionconv3x3
3x3 convolution with padding
defense/rs/archs/cifar_resnet.py:13
↓ 15 callersMethodget_momentum
(self, grad, momentum, **kwargs)
transferattack/advanced_objective/logit.py:105
↓ 15 callersMethodstep
(self, closure=None)
transferattack/model_related/awt.py:145
↓ 14 callersMethod__init__
(self, input_size, block, fwd_out, num_fwd, back_out, num_back, n, hard_mining = 0, loss_norm = False)
defense/hgd/inceptionresnet.py:478
↓ 14 callersMethod__init__
(self, input_size, block, fwd_out, num_fwd, back_out, num_back, n, hard_mining = 0, loss_norm = False)
defense/hgd/inception.py:454
↓ 14 callersFunctionconv_nd
Create a 1D, 2D, or 3D convolution module.
defense/diffpure/guided_diffusion/nn.py:30
↓ 14 callersMethodget_grad
Overridden for TIM attack.
transferattack/advanced_objective/cfm.py:73
↓ 13 callersFunctionsafe_divide
(a, b)
transferattack/model_related/ata_vit_utils/Transformer_Explainability/modules/layers_ours.py:10
↓ 13 callersMethodstate_dict
(self)
defense/diffpure/score_sde/models/ema.py:99
↓ 13 callersMethodupdate
(self, val, n=1)
transferattack/model_related/qaa_utils/train.py:174
↓ 11 callersFunction_cfg
(url='', **kwargs)
transferattack/model_related/setr_networks/t2t_vit.py:21
↓ 11 callersMethodget_loss
Calculate the loss
transferattack/input_transformation/mfi.py:167
↓ 10 callersFunction_cfg
(url='', **kwargs)
transferattack/model_related/setr_networks/swin_transformer.py:32
↓ 10 callersFunction_create_swin_transformer
(variant, pretrained=False, default_cfg=None, **kwargs)
transferattack/model_related/setr_networks/swin_transformer.py:553
↓ 10 callersFunction_resnet
( block: Type[Union[GhostBasicBlock, GhostBottleneck]], ghost_random_range: float, layers: List[in
transferattack/model_related/ghost_networks/resnet.py:302
↓ 10 callersMethoddrop
(self, data)
transferattack/advanced_objective/fft.py:119
↓ 9 callersMethod__init__
(self, channels, use_conv, dims=2, out_channels=None)
defense/diffpure/guided_diffusion/unet.py:99
↓ 9 callersFunctionget_current
()
defense/diffpure/guided_diffusion/logger.py:333
↓ 8 callersFunctiondefault_init
The same initialization used in DDPM.
defense/diffpure/score_sde/models/layers.py:88
↓ 8 callersMethodflush
Flush written text to both stdout and a file, if open.
defense/diffpure/utils.py:73
↓ 8 callersMethodlogkv_mean
(self, key, val)
defense/diffpure/guided_diffusion/logger.py:358
↓ 8 callersFunctionscore_fn
(x, t)
defense/diffpure/score_sde/models/utils.py:143
↓ 8 callersMethodshuffle
(self, x)
transferattack/input_transformation/bsr.py:57
↓ 8 callersMethodupdate
(self, val, n=1)
defense/at/lib/utils.py:22
↓ 7 callersMethod__init__
(self, num_features, bias=True)
defense/diffpure/score_sde/models/normalization.py:150
↓ 7 callersMethod__init__
(self, scale=1.0, noReLU=False)
defense/hgd/inceptionresnetv2.py:165
↓ 7 callersMethod__init__
(self, input_size, block, fwd_out, num_fwd, back_out, num_back, n, hard_mining = 0, loss_norm = False)
defense/hgd/resnext.py:205
↓ 7 callersMethod__init__
(self, net_type, input_size, block, fwd_out, num_fwd, back_out, num_back, n, hard_mining = 0, loss_norm = Fals
defense/hgd/resnet.py:258
↓ 7 callersMethod__init__
(self, model='inc_v3', ghost_keep_prob=0.994, ghost_random_range=0.16, *args, **kwargs)
transferattack/model_related/ghost.py:138
↓ 7 callersMethod__init__
(self)
transferattack/model_related/ghost_networks/inc_res_v2.py:73
↓ 7 callersFunctioncreate_model
Create the score model.
defense/diffpure/score_sde/models/utils.py:87
↓ 7 callersMethodcrop
(self, perturbation, img_width, img_height)
transferattack/generation/ge_advgan.py:155
↓ 7 callersMethodgradprop
(self, Z, X, S)
transferattack/model_related/ata_vit_utils/Transformer_Explainability/modules/layers_ours.py:41
↓ 7 callersMethodupdate
Update currently maintained parameters. Call this every time the parameters are updated, such as the result of the `optimizer.step()` ca
defense/diffpure/score_sde/models/ema.py:40
↓ 6 callersMethod__init__
(self, sde, score_fn, probability_flow=False)
defense/diffpure/score_sde/sampling.py:129
↓ 6 callersMethod__init__
(self, dim1, dim2, method='cat')
defense/diffpure/score_sde/models/layerspp.py:47
↓ 6 callersMethod__init__
(self, model_name='inc_v3', dhf_modules=None, mixup_weight_max=0.2, random_keep_prob=0.9, *args, **kwargs)
transferattack/model_related/dhf.py:236
↓ 6 callersMethod_compute_cond_module
(self, module, x)
defense/diffpure/score_sde/models/ncsnv2.py:381
↓ 6 callersMethodbackward
(ctx, grad_output)
transferattack/model_related/qaa_utils/archs/apot/quant_layer.py:115
↓ 6 callersMethodclear
(self)
transferattack/gradient/anda.py:207
↓ 6 callersMethodclose
(self)
defense/diffpure/guided_diffusion/logger.py:399
↓ 6 callersFunctionconv3x3
3x3 convolution with padding
transferattack/model_related/llta_networks/models/decayresnet.py:30
↓ 6 callersFunctioneval
(model, dataloader, is_targeted)
main.py:87
↓ 6 callersFunctionget_act
Get activation functions from the config file.
defense/diffpure/score_sde/models/layers.py:29
↓ 6 callersMethodget_loss
(self, mid_t_fmap, mid_s_fmap)
transferattack/advanced_objective/aa.py:76
↓ 6 callersFunctionlog
Write the sequence of args, with no separators, to the console and output files (if you've configured an output file).
defense/diffpure/guided_diffusion/logger.py:255
↓ 6 callersFunctionmean_flat
Take the mean over all non-batch dimensions.
defense/diffpure/guided_diffusion/nn.py:94
↓ 6 callersFunctionregister_recr
(net_, count, place_in_unet)
transferattack/generation/diffattack.py:513
↓ 6 callersFunctionreshape_heads_to_batch_dim
(tensor)
transferattack/generation/diffattack.py:473
↓ 6 callersMethodreverse
Create the reverse-time SDE/ODE. Args: score_fn: A time-dependent score-based model that takes x and t and returns the score. probabi
defense/diffpure/score_sde/sde_lib.py:79
↓ 5 callersMethod__init__
2D discrete wavelet transform (DWT) for 2D image decomposition :param wavename: pywt.wavelist(); in the paper, 'chx.y' denotes 'biorx
transferattack/model_related/metassa.py:363
↓ 5 callersMethod__init__
(self, model_name, epsilon=16/255, alpha=1.6/255, epoch=10, decay=1., num_style=20, gamma=0.5, beta = 2.0, tar
transferattack/input_transformation/stm.py:44
↓ 5 callersMethod_compute_cond_module
(self, module, x)
defense/diffpure/score_sde/models/ncsnv2.py:279
↓ 5 callersMethod_fia_loss_adv_only
EA loss on ADV half only: sum(adv * weights)/numel. fmap_2b: [2B, C, H, W], weights: [B, C, H, W]
transferattack/advanced_objective/mfaa.py:108
↓ 5 callersMethod_l2_normalize_per_sample
(t, eps: float = 1e-12)
transferattack/advanced_objective/mfaa.py:101
↓ 5 callersFunction_resnet
(arch, block, layers, pretrained, progress, **kwargs)
transferattack/model_related/qaa_utils/archs/apot/resnet.py:271
↓ 5 callersFunction_resnet
(arch, block, decayblock, layers, pretrained, progress, **kwargs)
transferattack/model_related/llta_networks/models/decayresnet.py:314
↓ 5 callersFunction_shape
(x, dim)
defense/diffpure/score_sde/models/up_or_down_sampling.py:199
↓ 5 callersFunctionconv1x1
1x1 convolution
transferattack/model_related/ghost_networks/resnet.py:59
↓ 5 callersFunctionconv1x1
1x1 convolution
transferattack/model_related/llta_networks/models/decayresnet.py:36
↓ 5 callersMethodconvert_to_fp16
Convert the torso of the model to float16.
defense/diffpure/guided_diffusion/unet.py:626
↓ 5 callersFunctionddpm_conv3x3
3x3 convolution with DDPM initialization.
defense/diffpure/score_sde/models/layers.py:118
↓ 5 callersFunctionlinear
Create a linear module.
defense/diffpure/guided_diffusion/nn.py:43
↓ 5 callersFunctionloss_fn
Compute the loss function. Args: model: A score model. batch: A mini-batch of training data. Returns: loss: A scalar that
defense/diffpure/score_sde/losses.py:73
↓ 5 callersFunctionmodel_and_diffusion_defaults
Defaults for image training.
defense/diffpure/guided_diffusion/script_util.py:51
↓ 5 callersFunctionmodel_fn
Compute the output of the score-based model. Args: x: A mini-batch of input data. labels: A mini-batch of conditioning va
defense/diffpure/score_sde/models/utils.py:107
↓ 5 callersFunctionnonlinearity
(x)
defense/diffpure/ddpm/unet_ddpm.py:35
↓ 4 callersFunctionNormalize
(in_channels)
defense/diffpure/ddpm/unet_ddpm.py:40
↓ 4 callersMethod__init__
(self, config)
defense/diffpure/ddpm/unet_ddpm.py:201
↓ 4 callersMethod__init__
(self, dim, input_resolution, depth, num_heads, window_size, mlp_ratio=4., qkv_bias=True, dro
transferattack/model_related/setr_networks/swin_transformer.py:378
next →1–100 of 2,264, ranked by callers