scenescoop
scenescoop copied to clipboard
error in using checkpoint (trained model)
Regardless of whether I use the version of TF specified in requirements file or if I upgrade to 1.10.0 , I get the error regarding the checkpoint (running on my mac locally). Any ideas on using a different pre-trained model?
NotFoundError (see above for traceback): Restoring from checkpoint failed. This is most likely due to a Variable name or other graph key that is missing from the checkpoint. Please ensure that you have not altered the graph expected based on the checkpoint. Original error:
Tensor name "lstm/basic_lstm_cell/bias" not found in checkpoint files /Users/silverbox/Dropbox/video_manipulation/scenescoop/models/im2txt/checkpoints/model.ckpt-3000000
[[Node: save/RestoreV2 = RestoreV2[dtypes=[DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_FLOAT, ..., DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_FLOAT, DT_FLOAT], _device="/job:localhost/replica:0/task:0/device:CPU:0"](_arg_save/Const_0_0, save/RestoreV2/tensor_names, save/RestoreV2/shape_and_slices)]]
Are you using the provided pre-trained model?
Yes, I tried using the one you provided as a Google drive link.
also having this issue. I think it has something to do with an update in the variable names used by tf (i.e. something like 'bias' vs 'biases')
update: was able to fix this issue by updating tf to 1.9 :-)
i got this error as well and was able to fix it by renaming the tensor names using the code found here: https://github.com/tensorflow/models/issues/466#issuecomment-391240675
Hello, I am also facing the same error. @genekogan can you please explain how did you go about solving the problem? Also @AaratiAkkapeddi can you please verify and confirm the tf version you upgraded to?