tfaddons0.10.0 package

Keras-based multi head attention layer

Neural Architecture Search (NAS) recurrent network cell.

Npairs multilabel loss

Pinball loss

Weighted cross-entropy loss for a sequence of logits.

Sigmoid focal crossentropy loss

Sparsemax loss

Triplet hard loss

Triplet semihard loss

Gelu

Hardshrink

Lisht

Mish

Rrelu

Softshrink

Sparsemax

Tanhshrink

Bahdanau Attention

Bahdanau Monotonic Attention

Implements Luong-style (multiplicative) attention scoring.

Monotonic attention mechanism with Luong-style energy function.

Monotonic attention

Attention Wrapper

Attention Wrapper State

Average Model Checkpoint

Time Stopping

TQDM Progress Bar

CRF binary score

CRF decode

CRF decode backward

CRF decode forward

CRF forward

CRF log likelihood

CRF log norm

CRF multitag sequence score

CRF sequence score

CRF unary score

Dynamic decode

An RNN Decoder abstract interface object.

Base Decoder

Basic Decoder

Basic decoder output

BeamSearch sampling decoder

Beam Search Decoder Output

Beam Search Decoder State

Final Beam Search Decoder Output

Factory function returning an optimizer class with decoupled weight de...

Gather tree

Gather tree from array

Hardmax

Adjust hsv in yiq

Angles to projective transforms

Blend

Compose transforms

Connected components

Cutout

Dense image warp

Equalize

Euclidean dist transform

Flat transforms to matrices

From 4D image

Get ndims

Interpolate bilinear

Interpolate spline

Matrices to flat transforms

Mean filter2d

Median filter2d

Random cutout

Random hsv in yiq

Resampler

Rotate

Sharpness

Shear x-axis

Shear y-axis

Sparse image warp

To 4D image

Transform

Translate

Translate xy dims

Translations to projective transforms

Uwrap

Wrap

Install TensorFlow SIG Addons

Gaussian Error Linear Unit

Correlation Cost Layer.

FilterResponseNormalization

Group normalization layer

Instance normalization layer

Maxout layer

LSTM cell with layer normalization and recurrent dropout.

Project into the Poincare ball with norm <= 1.0 - epsilon

Sparsemax activation function

Weight Normalization layer

Lookahead mechanism

Contrastive loss

Implements the GIoU loss function.

Hamming loss

Lifted structured loss

Npairs loss

Computes Kappa score between two raters

FBetaScore

Hamming distance

MatthewsCorrelationCoefficient

MultiLabelConfusionMatrix

RSquareThis is also called as coefficient of determination. It tells h...

F1Score

Conditional Gradient

Optimizer that implements the Adam algorithm with weight decay

Optimizer that implements the Momentum algorithm with weight_decay

Layer-wise Adaptive Moments

Lazy Adam

Moving Average

NovoGrad

Rectified Adam (a.k.a. RAdam)

Stochastic Weight Averaging

Yogi

Parse time

Objects exported from other packages

Register all

Register custom kernels

Register keras objects

Safe cumprod

Bernoulli sample

Categorical sample

Sampler

Base abstract class that allows the user to customize sampling.

Greedy Embedding Sampler

Inference Sampler

Sample Embedding Sampler

A training sampler that adds scheduled sampling

Scheduled Output Training Sampler

A Sampler for use during training.

Skip gram sample

Skip gram sample with text vocab

Version of TensorFlow SIG Addons

Tile batch

Viterbi decode