vm.torch.nn.functional

Functions

adaptive_avg_pool1d(…) : Apply the 1d adaptive average pooling to input.

adaptive_avg_pool2d(…) : Apply the 2d adaptive average pooling to input.

adaptive_avg_pool3d(…) : Apply the 3d adaptive average pooling to input.

adaptive_max_pool1d(…) : Apply the 1d adaptive max pooling to input.

adaptive_max_pool2d(…) : Apply the 2d adaptive max pooling to input.

adaptive_max_pool3d(…) : Apply the 3d adaptive max pooling to input.

avg_pool1d(…) : Apply the 1d average pooling to input.

avg_pool2d(…) : Apply the 2d average pooling to input.

avg_pool3d(…) : Apply the 3d average pooling to input.

batch_norm(…) : Apply the batch normalization to input. [Ioffe & Szegedy, 2015].

binary_cross_entropy_with_logits(…) : Compute the sigmoid cross entropy with contiguous target.

channel_shuffle(…) : Apply group shuffle to each channel of input. [Zhang et.al, 2017].

conv1d(…) : Apply the 1d convolution to input.

conv2d(…) : Apply the 2d convolution to input.

conv3d(…) : Apply the 3d convolution to input.

conv_transpose1d(…) : Apply the 1d deconvolution to input.

conv_transpose2d(…) : Apply the 2d deconvolution to input.

conv_transpose3d(…) : Apply the 3d deconvolution to input.

cross_entropy(…) : Compute the softmax cross entropy with sparse labels.

ctc_loss(…) : Compute the ctc loss with batched labels. [Graves & Gomez, 2006].

depthwise_conv2d(…) : Apply the 2d depthwise convolution to input. [Chollet, 2016].

drop_block2d(…) : Set the blocks over input to zero randomly. [Ghiasi et.al, 2018].

drop_path(…) : Set the examples over input to zero randomly. [Larsson et.al, 2016].

dropout(…) : Set the elements of input to zero randomly. [Srivastava et.al, 2014].

elu(…) : Apply the exponential linear unit to input. [Clevert et.al, 2015].

gelu(…) : Apply the gaussian error linear unit to input. [Hendrycks & Gimpel, 2016].

group_norm(…) : Apply the group normalization to input. [Wu & He, 2018].

hardsigmoid(…) : Apply the hard sigmoid function to input.

hardswish(…) : Apply the hard swish function to input. [Howard et.al, 2019].

kl_div(…) : Compute the Kullback-Leibler divergence.

l1_loss(…) : Compute the element-wise absolute value difference.

layer_norm(…) : Apply the layer normalization to input. [Ba et.al, 2016]

leaky_relu(…) : Apply the leaky rectified linear unit to input.

linear(…) : Apply the linear transformation to input.

local_response_norm(…) : Apply the local response normalization to input. [Krizhevsky et.al, 2012].

log_softmax(…) : Apply the composite of logarithm and softmax to input.

interpolate(…) : Resize input via interpolating neighborhoods.

max_pool1d(…) : Apply the 1d max pooling to input.

max_pool2d(…) : Apply the 2d max pooling to input.

max_pool3d(…) : Apply the 3d max pooling to input.

mse_loss(…) : Compute the element-wise squared error.

multi_head_attention_forward(…) : Apply the multihead attention to input. [Vaswani et.al, 2017].

nll_loss(…) : Compute the negative likelihood loss with sparse labels.

normalize(…) : Apply the \(L_{p}\) normalization to the input.

pad(…) : Pad the input according to the given sizes.

pixel_shuffle(…) : Rearrange depth elements of input into pixels.

pixel_unshuffle(…) : Rearrange pixels of input into depth elements.

prelu(…) : Apply the parametric rectified linear unit to input. [He et.al, 2015].

relu(…) : Apply rectified linear unit to input. [Nair & Hinton, 2010].

relu6(…) : Apply the clipped-6 rectified linear unit to input. [Krizhevsky, 2010].

selu(…) : Compute the sigmoid focal loss with sparse labels. [Lin et.al, 2017].

sigmoid(…) : Apply the sigmoid function to input.

sigmoid_focal_loss(…) : Compute the sigmoid focal loss with sparse labels. [Lin et.al, 2017].

silu(…) : Apply the sigmoid linear unit to input. [Hendrycks & Gimpel, 2016].

smooth_l1_loss(…) : Compute the element-wise error transited from L1 and L2. [Girshick, 2015].

softmax(…) : Apply the softmax function to input.

sync_batch_norm(…) : Apply the sync batch normalization to input. [Ioffe & Szegedy, 2015].

tanh(…) : Apply the tanh function to input.

unfold(…) : Extract the sliding blocks from input.

upsample(…) : Upsample input via interpolating neighborhoods.

upsample_bilinear(…) : Upsample input via bilinear interpolating.

upsample_nearest(…) : Upsample input via nearest interpolating.