deepmd.tf.env

deepmd.tf.env#

Module that sets tensorflow working environment and exports important constants.

Attributes#

`GLOBAL_CONFIG`
`GLOBAL_NP_FLOAT_PRECISION`
`SHARED_LIB_DIR`
`SHARED_LIB_MODULE`
`tf_py_version`
`EMBEDDING_NET_PATTERN`
`FITTING_NET_PATTERN`
`TYPE_EMBEDDING_PATTERN`
`ATTENTION_LAYER_PATTERN`
`TRANSFER_PATTERN`
`default_tf_session_config`
`MODEL_VERSION`
`TF_VERSION`
`op_module`
`op_grads_module`
`GLOBAL_TF_FLOAT_PRECISION`

Functions#

`reset_default_tf_session_config`(→ None)	Limit tensorflow session to CPU or not.
`global_cvt_2_tf_float`(→ tensorflow.compat.v1.Tensor)	Cast tensor to globally set TF precision.
`global_cvt_2_ener_float`(→ tensorflow.compat.v1.Tensor)	Cast tensor to globally set energy precision.

Module Contents#

deepmd.tf.env.GLOBAL_CONFIG[source]#

deepmd.tf.env.GLOBAL_NP_FLOAT_PRECISION[source]#

deepmd.tf.env.SHARED_LIB_DIR[source]#

deepmd.tf.env.SHARED_LIB_MODULE = 'lib'[source]#

deepmd.tf.env.tf_py_version[source]#

deepmd.tf.env.EMBEDDING_NET_PATTERN[source]#

deepmd.tf.env.FITTING_NET_PATTERN[source]#

deepmd.tf.env.TYPE_EMBEDDING_PATTERN[source]#

deepmd.tf.env.ATTENTION_LAYER_PATTERN[source]#

deepmd.tf.env.TRANSFER_PATTERN[source]#

deepmd.tf.env.default_tf_session_config[source]#

deepmd.tf.env.reset_default_tf_session_config(cpu_only: bool) → None[source]#

Limit tensorflow session to CPU or not.

Parameters:

cpu_onlybool: If enabled, no GPU device is visible to the TensorFlow Session.

deepmd.tf.env.MODEL_VERSION[source]#

deepmd.tf.env.TF_VERSION[source]#

deepmd.tf.env.op_module[source]#

deepmd.tf.env.op_grads_module[source]#

deepmd.tf.env.GLOBAL_TF_FLOAT_PRECISION[source]#

deepmd.tf.env.global_cvt_2_tf_float(xx: tensorflow.compat.v1.Tensor) → tensorflow.compat.v1.Tensor[source]#

Cast tensor to globally set TF precision.

Parameters:

xxtf.Tensor: input tensor

Returns:

tf.Tensor: output tensor cast to GLOBAL_TF_FLOAT_PRECISION

deepmd.tf.env.global_cvt_2_ener_float(xx: tensorflow.compat.v1.Tensor) → tensorflow.compat.v1.Tensor[source]#

Cast tensor to globally set energy precision.

Parameters:

xxtf.Tensor: input tensor

Returns:

tf.Tensor: output tensor cast to GLOBAL_ENER_FLOAT_PRECISION