/TensorFlow 2.4


Restores a tensor from checkpoint files.

Reads a tensor stored in one or several files. If there are several files (for instance because a tensor was saved as slices), file_pattern may contain wildcard symbols (* and ?) in the filename portion only, not in the directory portion.

If a file_pattern matches several files, preferred_shard can be used to hint in which file the requested tensor is likely to be found. This op will first open the file at index preferred_shard in the list of matching files and try to restore tensors from that file. Only if some tensors or tensor slices are not found in that first file, then the Op opens all the files. Setting preferred_shard to match the value passed as the shard input of a matching Save Op may speed up Restore. This attribute only affects performance, not correctness. The default value -1 means files are processed in order.

See also RestoreSlice.

file_pattern A Tensor of type string. Must have a single element. The pattern of the files from which we read the tensor.
tensor_name A Tensor of type string. Must have a single element. The name of the tensor to be restored.
dt A tf.DType. The type of the tensor to be restored.
preferred_shard An optional int. Defaults to -1. Index of file to open first if multiple files match file_pattern.
name A name for the operation (optional).
A Tensor of type dt.

© 2020 The TensorFlow Authors. All rights reserved.
Licensed under the Creative Commons Attribution License 3.0.
Code samples licensed under the Apache 2.0 License.