tffs's Introduction

TFFS

TensorFlow Filesystem - Access Tensors Differently

A funny way to access your tensorflow model's tensors.

Use this project to map your model into a filesystem. Then, access your tensors as if they were files, using your favorite UNIX commands.

tffs is implemented using Filesystem in Userspace (FUSE). It requires tensorflow and fusepy to be installed.

To learn more, read the accompanying blog post.

Usage

Create a model - out of the scope of this project :)
Mount your model so it'll be accessible through the filesystem:
```
python tffs.py --model PATH_TO_MODEL --mount MOUNT_POINT
```
PATH_TO_MODEL is either a directory containing a .meta file, or the .meta file itself.

If there's also a file containing the weights with the same name as the .meta file (without the .meta extension), it'll be loaded as well.
Reap the fruits. Assuming MOUNT_POINT is ~/tf:

Command	Description
`find ~/tf`	list all scopes and tensors
`find ~/tf -type f`	list all tensors
`xattr -l ~/tf/.../tensor`	get attributes of a tensor
`cat ~/tf/.../tensor`	print the value found in a tensor
`~/tf/bin/inputs -d 3 ~/tf/.../tensor`	print the inputs to a tensor, recursively
`~/tf/bin/outputs --no-fs .../tensor`	print the outputs to a tensor, without using the mount prefix

tffs's People

Stargazers

Watchers

tffs's Issues

No such file or directory

Hi @yoel-zeldes , what a great idea. I wanted to give it a try so I cloned your repo, installed fusepy but I'm getting the following error when I try python ../tffs/tffs.py --model chkpts/f64c32_conv/chkpt-682.meta --mount mount/

Traceback (most recent call last):
  File "../tffs/tffs.py", line 290, in <module>
    FUSE(TfFs(mount_point=args.mount, model_path=args.model), args.mount, foreground=True)
  File "../tffs/tffs.py", line 162, in __init__
    self._populate_bin(mount_point)
  File "../tffs/tffs.py", line 186, in _populate_bin
    TENSOR_TO_TENSORS_PLACEHOLDER=str(tensor_to_tensors()))
  File "../tffs/tffs.py", line 145, in _read_template_script
    with open(template_path, 'r') as f:
IOError: [Errno 2] No such file or directory: 'print_tensor_dependencies.py'

ls -1 chkpts/f64c32_conv/chkpt-682.*
chkpts/f64c32_conv/chkpt-682.data-00000-of-00001
chkpts/f64c32_conv/chkpt-682.index
chkpts/f64c32_conv/chkpt-682.meta

What am I doing wrong here?

My conda environment has:

Package                            Version
---------------------------------- -----------
absl-py                            0.5.0
astor                              0.7.1
backports-abc                      0.5
backports.functools-lru-cache      1.5
backports.shutil-get-terminal-size 1.0.0
backports.weakref                  1.0.post1
bashplotlib                        0.6.5
certifi                            2018.10.15
chardet                            3.0.4
configparser                       3.5.0
cycler                             0.10.0
decorator                          4.3.0
enum34                             1.1.6
funcsigs                           1.0.2
functools32                        3.2.3.post2
fusepy                             3.0.1
futures                            3.2.0
gast                               0.2.0
grpcio                             1.12.1
h5py                               2.8.0
idna                               2.7
influxdb                           5.2.0
ipython                            5.8.0
ipython-genutils                   0.2.0
kiwisolver                         1.0.1
linecache2                         1.0.0
Markdown                           3.0.1
matplotlib                         2.2.3
mkl-fft                            1.0.6
mkl-random                         1.0.1
mock                               2.0.0
numpy                              1.15.3
olefile                            0.46
pathlib2                           2.3.2
pbr                                4.3.0
pexpect                            4.6.0
pickleshare                        0.7.5
Pillow                             5.3.0
pip                                18.1
progressbar                        2.5
prompt-toolkit                     1.0.15
protobuf                           3.6.0
ptyprocess                         0.6.0
pudb                               2018.1
Pygments                           2.2.0
pyparsing                          2.2.2
python-dateutil                    2.7.5
pytz                               2018.7
requests                           2.20.0
scandir                            1.9.0
setuptools                         40.4.3
simplegeneric                      0.8.1
singledispatch                     3.4.0.3
six                                1.11.0
subprocess32                       3.5.3
tabulate                           0.8.2
tensorboard                        1.10.0
tensorflow                         1.10.0
termcolor                          1.1.0
tornado                            5.1.1
tqdm                               4.28.1
traceback2                         1.4.0
traitlets                          4.3.2
unittest2                          1.1.0
urllib3                            1.24
urwid                              2.0.1
wcwidth                            0.1.7
Werkzeug                           0.14.1
wheel                              0.32.2

Recommend Projects