ValueError: Couldn't find 'checkpoint' file or checkpoints in given directory with Google's image classification built-in algorithm

Question

I've tried to fine-tune a resnet50 model following "Getting started with the built-in image classification algorithm" tutorial. I have stored resnet 50 imagenet checkpoints (found here) in my google cloud storage (path is gs://my-bucket/classification/models/resnet-50). It contains the following:

model.ckpt-0090.index
model.ckpt-0090.data-00000-of-00002
model.ckpt-0090.data-00001-of-00002

When I submit the training job, I specify with argument pretrained_checkpoint_path the above path. When job starts, I get the following error:

ValueError: Couldn't find 'checkpoint' file or checkpoints in given directory gs://my-bucket/classification/models/resnet-50

Should the name of the checkpoint files conform with some predefined format?

EDIT: actually, when uncompressed the checkpoints directory also contained the same 3 files as above but prefixed with _.. I got the error so I removed these prefixed files but with no more success.