BatchToValueDistribution(distribution, ndims) |
Distribution that converts the last few batch_ndims into values_ndims. |
Bernoulli(logits[, dtype]) |
Univariate Bernoulli distribution. |
Categorical(logits[, dtype]) |
Univariate Categorical distribution. |
Concrete(temperature, logits[, …]) |
The class of Concrete (or Gumbel-Softmax) distribution from (Maddison, 2016; Jang, 2016), served as the continuous relaxation of the OnehotCategorical. |
Discrete |
alias of tfsnippet.distributions.univariate.Categorical |
DiscretizedLogistic(mean, log_scale, bin_size) |
Discretized logistic distribution (Kingma et. |
Distribution(dtype, is_continuous, …) |
Base class for probability distributions. |
ExpConcrete(temperature, logits[, …]) |
The class of ExpConcrete distribution from (Maddison, 2016), transformed from Concrete by taking logarithm. |
FlowDistribution(distribution, flow) |
Transform a Distribution by a BaseFlow, as a new distribution. |
FlowDistributionDerivedTensor(tensor, …) |
A combination of a FlowDistribution derived tensor, and its original stochastic tensor from the base distribution. |
Mixture(categorical, components[, …]) |
Mixture distribution. |
Normal(mean[, std, logstd, …]) |
Univariate Normal distribution. |
OnehotCategorical(logits[, dtype]) |
One-hot multivariate Categorical distribution. |
Uniform([minval, maxval, …]) |
Univariate Uniform distribution. |
AnnealingVariable(name, initial_value, ratio) |
A non-trainable tf.Variable, whose value will be annealed as training goes by. |
CheckpointSavableObject |
Base class for all objects that can be saved via CheckpointSaver. |
CheckpointSaver(variables, save_dir[, …]) |
Save and restore tf.Variable, ScheduledVariable and CheckpointSavableObject with tf.train.Saver. |
DefaultMetricFormatter |
Default training metric formatter. |
EventKeys |
Defines event keys for TFSnippet. |
MetricFormatter |
Base class for a training metrics formatter. |
MetricLogger([summary_writer, …]) |
Logger for the training metrics. |
ScheduledVariable(name, initial_value[, …]) |
A non-trainable tf.Variable, whose value might need to be changed as training goes by. |
TrainLoop(param_vars[, var_groups, …]) |
Training loop object. |
AnnealingScalar(loop, initial_value, ratio) |
A DynamicValue scalar, which anneals every few epochs or steps. |
BaseTrainer(loop[, ensure_variables_initialized]) |
Base class for all trainers. |
DynamicValue |
Dynamic values to be fed into trainers and evaluators. |
Evaluator(loop, metrics, inputs, data_flow) |
Class to compute evaluation metrics. |
LossTrainer(**kwargs) |
A subclass of BaseTrainer, which optimizes a single loss. |
Trainer(loop, train_op, inputs, data_flow[, …]) |
A subclass of BaseTrainer, executing a training operation per step. |
Validator(**kwargs) |
Class to compute validation loss and other metrics. |
VariationalChain(variational, model[, …]) |
Chain of the variational and model nets for variational inference. |
VariationalEvaluation(vi) |
Factory for variational evaluation outputs. |
VariationalInference(log_joint, latent_log_probs) |
Class for variational inference. |
VariationalLowerBounds(vi) |
Factory for variational lower-bounds. |
VariationalTrainingObjectives(vi) |
Factory for variational training objectives. |
BayesianNet([observed]) |
Bayesian networks. |
DataFlow |
Data flows are objects for constructing mini-batch iterators. |
DataMapper |
Base class for all data mappers. |
SlidingWindow(data_array, window_size) |
DataMapper for producing sliding windows according to indices. |
Config |
Base class for defining config values. |
ConfigField(type[, default, description, …]) |
A config field. |
GraphKeys |
Defines TensorFlow graph collection keys for TFSnippet. |
InvertibleMatrix(size[, strict, dtype, …]) |
A matrix initialized to be an invertible, orthogonal matrix. |
VarScopeObject([name, scope]) |
Base class for objects that own a variable scope. |
SummaryCollector([collections, …]) |
Collecting summaries and histograms added by tfsnippet.add_summary() and tfsnippet.add_histogram(). |
StochasticTensor(distribution, tensor[, …]) |
Samples or observations of a stochastic variable. |
