deep-fry / mayo Goto Github PK

First thank you very much for providing such a convenient tool .
I am tring to test the base example according to the README.md by run following scripts:
./my models/lenet5.yaml datasets/mnist.yaml system.checkpoint.load=pretrained eval

However, I got following error report:

Traceback (most recent call last):
  File "./my.py", line 13, in <module>
    cli.CLI().main()
  File "mayo/cli.py", line 300, in main
    commands[each]()
  File "mayo/cli.py", line 210, in cli_eval
    return self._get_session('validate').eval()
  File "/home/daisurong/workspace_2021/mayo/mayo/session/base.py", line 50, in wrapped
    return func(self, *args, **kwargs)
  File "/home/daisurong/workspace_2021/mayo/mayo/session/eval.py", line 19, in eval
    self.load_checkpoint(key)
  File "/home/daisurong/workspace_2021/mayo/mayo/session/base.py", line 50, in wrapped
    return func(self, *args, **kwargs)
  File "/home/daisurong/workspace_2021/mayo/mayo/session/base.py", line 224, in load_checkpoint
    restore_vars = self.checkpoint.load(name)
  File "/home/daisurong/workspace_2021/mayo/mayo/session/checkpoint.py", line 102, in load
    reader = tf.train.NewCheckpointReader(path)
  File "/home/daisurong/.local/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 636, in NewCheckpointReader
    return CheckpointReader(compat.as_bytes(filepattern))
  File "/home/daisurong/.local/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py", line 648, in __init__
    this = _pywrap_tensorflow_internal.new_CheckpointReader(filename)
DataLossError: not an sstable (bad magic number)

> /home/daisurong/.local/lib/python3.6/site-packages/tensorflow/python/pywrap_tensorflow_internal.py(648)__init__()
    647     def __init__(self, filename):
--> 648         this = _pywrap_tensorflow_internal.new_CheckpointReader(filename)
    649         try:

ipdb> i
*** NameError: name 'i' is not defined
ipdb> s  

Then, the program exited itself.
Could you please give me some tips on how to make it work correctly.
Thank you so much!

Hello, Is there any documentation to use dns pruning from scratch ?
I am working on training a model but i can't apply dns in it as it is not clear in the documentation

The model download link https://universityofcambridgecloud-my.sharepoint.com/:f:/g/personal/yaz21_cam_ac_uk/Es4FrWNmJe1ImBgR_T1PyoUBB1-D_UCmmeo6KCT_RHDsyQ?e=BzpTzk has been disabled. Is there any new link? Thanks.

thanks to giving the Mayo and i first feel think it will be a pretty tool
but i try to run your example .the FBS example.
i couldn't found the checkpoints of gate50
the message is as follows:
CheckpointNotFoundError: Checkpoint 'checkpoints/cifarnet/cifar10/gate50' not found.
could you offer the chepoints to me . i want to test it .thanks

Hi,

I think I found an issue with the accounting of MAC operations/# weights read in the memory (MEM) when using the dynamic channel pruning (DCP) feature. The number of MAC/MEM reported for gate.density values < 1 are more than they should be. This happens when a conv layer follows a non convolution layer like dropout. In case of density values of < 1, there is sparsity benefit from both input and output sides. However, when a conv follows a dropout layer, input side sparsity is not accounted for, even though the input to dropout is a sparse tensor from a conv layer.

Take the CIFARNET example with gate density values of 1.0 and 0.1. The results from the framework are pasted below. The errors are in the estimate of conv4 and conv7. For GD of 0.1, the estimates for conv4 and conv7 are 10x higher than they should be because of the above mentioned issue.

Gate Density = 1.0

And Gate Density =0.1

The pickle file shows "before" and "after" quantization values but the checkpoints only save the float32 model. Is there anyway to get the model with the quantized values themselves?