Commit e2f549e6 authored by Brennan Saeta's avatar Brennan Saeta Committed by TensorFlower Gardener
Browse files

Add buffering to RecordReader, and reader datasets.

Currently RecordReader reads from random access files. When reading from
high-latency data stores (such as GCS), it's often preferable to buffer reads.
This change adds buffering support to the reader datasets, and passes the
configurations down to their respective backend implementations.

PiperOrigin-RevId: 166245223
parent 84138ef8
Loading
Loading
Loading
Loading
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please to comment