LSTM Hyperparameter Tuning producing Checkpoint errors from access denied



error3
My temp folder permissions are all allowed and other files from this script are written there. I’ve disabled my firewall and virus scanner but no dice. Running as admin and have a lot of disk space. Tensorflow also at most up to date version…

What do? This is unfortunate as the LSTM model is producing the best hyperparameters compared to the other models for my data.

@Duncan,

Welcome to the Tensorflow Forum,

Sometimes it happens if the file is currently in use by another program. Then you can close the program or wait for it to finish using the file before attempting to restore the checkpoint.

I suggest, please update stack trace in text format instead of images.

Thank you!

Hi there, thanks for checking in.

The file is a checkpoint file that is being created when running the script; it does not exist before so cannot be being used by another program.
2023-07-20 14:49:35.235233: E tensorflow/tsl/platform/windows/windows_file_system.cc:363] ERROR: GetSymbolicLinkTarget cannot open file for \?\C:\Users\l1lba\AppData\Local\Temp\tmpb3xeohfz GetLastError: 5

2023-07-20 14:49:35.235643: W tensorflow/core/util/tensor_slice_reader.cc:97] Could not open C:\Users\l1lba\AppData\Local\Temp\tmpb3xeohfz: UNKNOWN: NewRandomAccessFile failed to Create/Open: C:\Users\l1lba\AppData\Local\Temp\tmpb3xeohfz : Access is denied.
; Input/output error
WARNING:tensorflow:Detecting that an object or model or tf.train.Checkpoint is being deleted with unrestored values. See the following logs for the specific values in question. To silence these warnings, use status.expect_partial(). See tf.train.Checkpoint  |  TensorFlow v2.16.1 details about the status object returned by the restore function.
WARNING:tensorflow:Value in checkpoint could not be found in the restored object: (root).keras_api.metrics.0.total
WARNING:tensorflow:Value in checkpoint could not be found in the restored object: (root).keras_api.metrics.0.count

@Duncan,

Thank you for sharing the stack trace.

To debug the issue further, could you please share standalone code to reproduce the issue?