No checkpoint found, despite running prevously
drdeaton opened this issue · comments
Drew commented
Whenever I run train.py
, I get the following message:
Trying to restore saved checkpoints from ./logdir/train/2018-07-24T21-45-30 ... No checkpoint found.
files length: 444
It then continues, starting anew from step 0
How can I get it to continue where it left off?
Also, it's probably worth mentioning that I get this whenever it saves a checkpoint:
Storing checkpoint to ./logdir/train/2018-07-24T21-45-30 ...WARNING:tensorflow:Issue encountered when serializing trainable_variables.
Type is unsupported, or the types of the items don't match field type in CollectionDef. Note this is a warning and probably safe to ignore.
'filter_bias' has type str, but expected one of: int, long, bool
WARNING:tensorflow:Issue encountered when serializing variables.
Type is unsupported, or the types of the items don't match field type in CollectionDef. Note this is a warning and probably safe to ignore.
'filter_bias' has type str, but expected one of: int, long, bool
Done.
I don't think that is the issue, because when opening the checkpoint file in gedit, it seems like a valid cfg file format, but this could be to blame.
Yaque commented
i do that don't have this problem, now i train it 3500 step.
solmn1 commented
@Dysproh did you solve the problem?
Crystal Dias commented
@Dysproh Has a solution been found for this problem?
Kaylon commented
I am also running into this problem and I'm surprised this hasn't been fixed as well