MCPcopy Create free account

hub / github.com/HazyResearch/butterfly / types & classes

Types & classes260 in github.com/HazyResearch/butterfly

⨍ Functions 1,242◇ Types & classes 260
↓ 36 callersClassButterflyProduct
Product of butterfly matrices. The order are chosen by softmaxes, which are learnable.
learning_transforms/butterfly_old.py:162
↓ 28 callersClassButterfly
Product of log N butterfly factors, each is a block 2x2 of diagonal matrices. Compatible with torch.nn.Linear. Parameters: in_size: s
torch_butterfly/butterfly.py:15
↓ 27 callersClassFixedPermutation
torch_butterfly/permutation.py:55
↓ 25 callersClassButterfly
Butterfly matrix of size n x n where only the diagonal and the k-th subdiagonal and superdiagonal are nonzero.
learning_transforms/butterfly_old.py:31
↓ 22 callersClassBlock2x2DiagProduct
Product of block 2x2 diagonal matrices.
learning_transforms/butterfly_old.py:265
↓ 20 callersClassReal2Complex
torch_butterfly/complex_utils.py:115
↓ 18 callersClassComplex2Real
torch_butterfly/complex_utils.py:120
↓ 16 callersClassFire
cnn/models/squeezenet.py:16
↓ 12 callersClassButterflyConv2d
Product of log N butterfly factors, each is a block 2x2 of diagonal matrices. Parameters: in_channels: size of input out_channels
cnn/models/butterfly_conv.py:29
↓ 11 callersClassButterfly
Product of log N butterfly factors, each is a block 2x2 of diagonal matrices. Compatible with torch.nn.Linear. Parameters: in_size: s
butterfly/butterfly.py:13
↓ 11 callersClassTensorProduct
torch_butterfly/combine.py:80
↓ 10 callersClassDiagonal
torch_butterfly/diagonal.py:11
↓ 9 callersClassAverageMeter
Computes and stores the average and current value
cnn/imagenet_experiment.py:398
↓ 9 callersClassInception
cnn/models/googlenet.py:7
↓ 9 callersClassLowRankConv2d
cnn/models/low_rank_conv.py:8
↓ 8 callersClassAverageMeter
Computes and stores the average and current value
cnn/imagenet_amp.py:417
↓ 7 callersClassBlockPermProduct
Product of block permutation matrices.
learning_transforms/butterfly_old.py:596
↓ 6 callersClassButterflyBmm
Product of log N butterfly factors, each is a block 2x2 of diagonal matrices. Perform batch matrix multiply. Parameters: in_size: siz
butterfly/butterfly.py:313
↓ 6 callersClassDiagonalMultiplySum
torch_butterfly/special.py:417
↓ 6 callersClassMobileNet
cnn/mobilenet_imagenet.py:98
↓ 6 callersClassPermutation
Product of log N permutation factors. Parameters: size: size of input (and of output) share_logit: whether the logits in the perm
butterfly/permutation.py:9
↓ 6 callersClassResNetOriginal
cnn/models/resnet_original.py:86
↓ 5 callersClassButterfly1x1Conv
Product of log N butterfly factors, each is a block 2x2 of diagonal matrices.
cnn/mobilenet_imagenet.py:40
↓ 5 callersClassDenseNet
cnn/models/densenet.py:36
↓ 5 callersClassPResNet
cnn/models/presnet.py:112
↓ 5 callersClassPreActResNet
cnn/models/preact_resnet.py:65
↓ 5 callersClassResNet
cnn/models/resnet.py:99
↓ 4 callersClassBlock2x2Diag
Block matrix of size n x n of the form [[A, B], [C, D]] where each of A, B, C, D are diagonal. This means that only the diagonal and the n//2-th
learning_transforms/butterfly_old.py:203
↓ 4 callersClassKOP2d
convolution/models/kops.py:15
↓ 4 callersClassModelAndLoss
cnn/imagenet/training.py:30
↓ 4 callersClassResNeXt
cnn/models/resnext.py:40
↓ 4 callersClassSepConv
Separable Convolution.
cnn/models/pnasnet.py:10
↓ 3 callersClassHstackDiagProduct
Product of HstackDiag matrices.
learning_transforms/hstack_diag.py:45
↓ 3 callersClassTransition
cnn/models/densenet.py:24
↓ 2 callersClassBlock2x2DiagRectangular
Block matrix of size k n x k n of the form [[A, B], [C, D]] where each of A, B, C, D are diagonal. This means that only the diagonal and the n//2-
learning_transforms/butterfly_old.py:344
↓ 2 callersClassBlockPerm
Block permutation matrix of size n x n.
learning_transforms/butterfly_old.py:533
↓ 2 callersClassButterflyConv2dBBT
Product of log N butterfly factors, each is a block 2x2 of diagonal matrices. Parameters: in_channels: size of input out_channels
cnn/models/butterfly_conv.py:104
↓ 2 callersClassButterflyUnitary
Same as Butterfly, but constrained to be unitary Compatible with torch.nn.Linear. Parameters: in_size: size of input out_size
torch_butterfly/butterfly.py:210
↓ 2 callersClassDALIWrapper
cnn/imagenet/dataloaders.py:103
↓ 2 callersClassDPN
cnn/models/dpn.py:38
↓ 2 callersClassFixedPermutation
butterfly/permutation.py:59
↓ 2 callersClassFixedPermutation
learning_transforms/butterfly_old.py:641
↓ 2 callersClassFlatten
cnn/models/layers.py:18
↓ 2 callersClassHadamard1x1Conv
cnn/shufflenet_imagenet.py:59
↓ 2 callersClassKnowledgeDistillationLoss
Loss with knowledge distillation.
cnn/imagenet/smoothing.py:29
↓ 2 callersClassLabelSmoothing
NLL loss with label smoothing.
cnn/imagenet/smoothing.py:5
↓ 2 callersClassMixUpWrapper
cnn/imagenet/mixup.py:18
↓ 2 callersClassNLLMultiLabelSmooth
cnn/imagenet/mixup.py:33
↓ 2 callersClassPNASNet
cnn/models/pnasnet.py:71
↓ 2 callersClassPermutationFactor
A single permutation factor. Parameters: size: size of input (and of output)
butterfly/permutation.py:79
↓ 2 callersClassPrefetchedWrapper
cnn/imagenet/dataloaders.py:190
↓ 2 callersClassShuffleBlock
cnn/models/shufflenetv2.py:10
↓ 2 callersClassShuffleNet
cnn/shufflenet_imagenet.py:143
↓ 2 callersClassShuffleNet
cnn/models/shufflenet.py:51
↓ 2 callersClassSqueezeNet
cnn/models/squeezenet.py:39
↓ 2 callersClassWide_ResNet
cnn/models/wide_resnet.py:70
↓ 2 callersClassdata_prefetcher
cnn/imagenet_experiment.py:239
↓ 1 callersClassAverageMeter
Computes and stores the average and current value
cnn/teacher_covariance.py:167
↓ 1 callersClassAverageMeter
Computes and stores the average and current value
cnn/train_utils.py:22
↓ 1 callersClassBasicBlock
cnn/models/shufflenetv2.py:32
↓ 1 callersClassBlock
Depthwise conv + Pointwise conv
cnn/mobilenet_imagenet.py:57
↓ 1 callersClassBlock
Grouped convolution block.
cnn/models/resnext.py:10
↓ 1 callersClassBlock
Depthwise conv + Pointwise conv
cnn/models/mobilenet.py:17
↓ 1 callersClassBlock
expand + depthwise + pointwise
cnn/models/mobilenetv2.py:11
↓ 1 callersClassBlock2x2DiagBmm
Block matrix of size n x n of the form [[A, B], [C, D]] where each of A, B, C, D are diagonal. This means that only the diagonal and the n//2-th
learning_transforms/butterfly_old.py:460
↓ 1 callersClassBottleneck
cnn/shufflenet_imagenet.py:86
↓ 1 callersClassBottleneck
cnn/models/dpn.py:7
↓ 1 callersClassBottleneck
cnn/models/shufflenet.py:22
↓ 1 callersClassButterflyBmm
Same as Butterfly, but performs batched matrix multiply. Compatible with torch.nn.Linear. Parameters: in_size: size of input
torch_butterfly/butterfly.py:311
↓ 1 callersClassCirculant1x1Conv
cnn/models/circulant1x1conv.py:32
↓ 1 callersClassDownBlock
cnn/models/shufflenetv2.py:58
↓ 1 callersClassFeatures
gumbel-sinkhorn/my_sorting_model.py:6
↓ 1 callersClassGoogLeNet
cnn/models/googlenet.py:56
↓ 1 callersClassHstackDiag
Horizontally stacked diagonal matrices of size n x 2n. Each entry in a 2x2 matrix of polynomials.
learning_transforms/hstack_diag.py:14
↓ 1 callersClassHybridTrainPipe
cnn/imagenet/dataloaders.py:19
↓ 1 callersClassHybridValPipe
cnn/imagenet/dataloaders.py:69
↓ 1 callersClassLambdaLayer
convolution/models/resnet_cifar.py:21
↓ 1 callersClassLambdaLayer
cnn/models/resnet_original.py:45
↓ 1 callersClassMobileNet
cnn/models/mobilenet.py:50
↓ 1 callersClassMobileNetV2
cnn/models/mobilenetv2.py:40
↓ 1 callersClassNode
torch_butterfly/permutation.py:169
↓ 1 callersClassResNet
cnn/models/resnet_imagenet.py:194
↓ 1 callersClassResNet18
convolution/models/resnet.py:110
↓ 1 callersClassResNetBuilder
cnn/imagenet/resnet.py:10
↓ 1 callersClassSENet
cnn/models/senet.py:79
↓ 1 callersClassShuffleBlock
cnn/shufflenet_imagenet.py:74
↓ 1 callersClassShuffleBlock
cnn/models/shufflenet.py:10
↓ 1 callersClassShuffleNetV2
cnn/models/shufflenetv2.py:96
↓ 1 callersClassSplitBlock
cnn/models/shufflenetv2.py:22
↓ 1 callersClassStderrTee
convolution/tee.py:113
↓ 1 callersClassStdoutTee
convolution/tee.py:106
↓ 1 callersClassTensorPermutation
cnn/models/presnet.py:207
↓ 1 callersClassToeplitzlike1x1Conv
cnn/models/toeplitzlike1x1conv.py:163
↓ 1 callersClassTuneReportCheckpointCallback
convolution/ray_runner.py:80
↓ 1 callersClassVGG
cnn/models/vgg.py:14
↓ 1 callersClassdata_prefetcher
cnn/train_utils.py:39
ClassAdaptiveConcatPool2d
cnn/models/layers.py:6
ClassAlexNet
cnn/models/lenet.py:88
ClassAverageMeter
cnn/imagenet/logger.py:40
ClassBasicBlock
convolution/models/resnet_cifar.py:30
next →1–100 of 260, ranked by callers