deepmd.tf.nvnmd.utils.network

Attributes

log

Functions

get_sess()
matmul2_qq(a, b, nbit)	Quantized matmul operation for 2d tensor.
matmul3_qq(a, b, nbit)	Quantized matmul operation for 3d tensor.
qf(x, nbit)	Quantize and floor tensor x with quantification precision nbit.
qr(x, nbit)	Quantize and round tensor x with quantification precision nbit.
tanh4(x)
one_layer_wb(shape, outputs_size, bavg, stddev, ...)
one_layer_t(shape, outputs_size, bavg, stddev, ...)
one_layer(inputs, outputs_size[, activation_fn, ...])	Build one layer with continuous or quantized value.

Module Contents

deepmd.tf.nvnmd.utils.network.log

deepmd.tf.nvnmd.utils.network.get_sess()

deepmd.tf.nvnmd.utils.network.matmul2_qq(a, b, nbit)

Quantized matmul operation for 2d tensor. a and b is input tensor, nbit represent quantification precision.

deepmd.tf.nvnmd.utils.network.matmul3_qq(a, b, nbit)

Quantized matmul operation for 3d tensor. a and b is input tensor, nbit represent quantification precision.

deepmd.tf.nvnmd.utils.network.qf(x, nbit)

Quantize and floor tensor x with quantification precision nbit.

deepmd.tf.nvnmd.utils.network.qr(x, nbit)

Quantize and round tensor x with quantification precision nbit.

deepmd.tf.nvnmd.utils.network.tanh4(x)

deepmd.tf.nvnmd.utils.network.one_layer_wb(shape, outputs_size, bavg, stddev, precision, trainable, initial_variables, seed, uniform_seed, name)

deepmd.tf.nvnmd.utils.network.one_layer_t(shape, outputs_size, bavg, stddev, precision, trainable, initial_variables, seed, uniform_seed, name)

deepmd.tf.nvnmd.utils.network.one_layer(inputs, outputs_size, activation_fn=tf.nn.tanh, precision=GLOBAL_TF_FLOAT_PRECISION, stddev=1.0, bavg=0.0, name='linear', reuse=None, seed=None, use_timestep=False, trainable=True, useBN=False, uniform_seed=False, initial_variables=None, mixed_prec=None, final_layer=False)

Build one layer with continuous or quantized value. Its weight and bias can be initialed with random or constant value.