gheinrich/digits-gan

DIGITS

DIGITS (the Deep Learning GPU Training System) is a webapp for training deep learning models.

Installation

Installation method	Supported platform[s]	Available versions	Instructions
Deb packages	Ubuntu 14.04	14.04 repo	docs/UbuntuInstall.md
Docker	Linux	DockerHub tags	nvidia-docker wiki
Source	Ubuntu 14.04, 16.04	GitHub tags	docs/BuildDigits.md

Usage

Once you have installed DIGITS, visit docs/GettingStarted.md for an introductory walkthrough.

Then, take a look at some of the other documentation at docs/ and examples/:

• Getting started with Torch
• Fine-tune a pretrained model
• Train an autoencoder network
• Train a regression network
• Train a Siamese network
• Train a text classification network
• Train an object detection network
• Learn more about weight initialization
• Use Python layers in your Caffe networks
• Download a model and use it to classify an image outside of DIGITS
• Overview of the REST API

Get help

Installation issues

• First, check out the instructions above
• Then, ask questions on our user group

Usage questions

• First, check out the Getting Started page
• Then, ask questions on our user group

Bugs and feature requests

• Please let us know by filing a new issue
• Bonus points if you want to contribute by opening a pull request!
  • You will need to send a signed copy of the Contributor License Agreement to [email protected] before your change can be accepted.

digits-gan's Issues

Tensorflow version

Hello,

What TF version matches with this implementation?
I ask this because I run the MNIST+LeNet tutorial in TF instead of caffe and I got some errors.

It is what I did:

$ pip install https://ci.tensorflow.org/view/Nightly/job/nightly-matrix-linux-gpu/TF_BUILD_IS_OPT=OPT,TF_BUILD_IS_PIP=PIP,TF_BUILD_PYTHON_VERSION=PYTHON2,label=gpu-linux/lastSuccessfulBuild/artifact/pip_test/whl/tensorflow_gpu-1.1.0-cp27-none-linux_x86_64.whl

import tensorflow

Ok. Then,

$ pip install -e $DIGITS_ROOT

At this point I'm not able to run without getting an error.

import tensorflow
/usr/local/lib/python2.7/dist-packages/tensorflow/__init__.py in <module>()
     22 
     23 # pylint: disable=wildcard-import
---> 24 from tensorflow.python import *
     25 # pylint: enable=wildcard-import
     26 

/usr/local/lib/python2.7/dist-packages/tensorflow/python/__init__.py in <module>()
     50 
     51 # Protocol buffers
---> 52 from tensorflow.core.framework.graph_pb2 import *
     53 from tensorflow.core.framework.node_def_pb2 import *
     54 from tensorflow.core.framework.summary_pb2 import *

/usr/local/lib/python2.7/dist-packages/tensorflow/core/framework/graph_pb2.py in <module>()
     14 
     15 
---> 16 from tensorflow.core.framework import node_def_pb2 as tensorflow_dot_core_dot_framework_dot_node__def__pb2
     17 from tensorflow.core.framework import function_pb2 as tensorflow_dot_core_dot_framework_dot_function__pb2
     18 from tensorflow.core.framework import versions_pb2 as tensorflow_dot_core_dot_framework_dot_versions__pb2

/usr/local/lib/python2.7/dist-packages/tensorflow/core/framework/node_def_pb2.py in <module>()
     14 
     15 
---> 16 from tensorflow.core.framework import attr_value_pb2 as tensorflow_dot_core_dot_framework_dot_attr__value__pb2
     17 
     18 

/usr/local/lib/python2.7/dist-packages/tensorflow/core/framework/attr_value_pb2.py in <module>()
     14 
     15 
---> 16 from tensorflow.core.framework import tensor_pb2 as tensorflow_dot_core_dot_framework_dot_tensor__pb2
     17 from tensorflow.core.framework import tensor_shape_pb2 as tensorflow_dot_core_dot_framework_dot_tensor__shape__pb2
     18 from tensorflow.core.framework import types_pb2 as tensorflow_dot_core_dot_framework_dot_types__pb2

/usr/local/lib/python2.7/dist-packages/tensorflow/core/framework/tensor_pb2.py in <module>()
     14 
     15 
---> 16 from tensorflow.core.framework import resource_handle_pb2 as tensorflow_dot_core_dot_framework_dot_resource__handle__pb2
     17 from tensorflow.core.framework import tensor_shape_pb2 as tensorflow_dot_core_dot_framework_dot_tensor__shape__pb2
     18 from tensorflow.core.framework import types_pb2 as tensorflow_dot_core_dot_framework_dot_types__pb2

/usr/local/lib/python2.7/dist-packages/tensorflow/core/framework/resource_handle_pb2.py in <module>()
     20   package='tensorflow',
     21   syntax='proto3',
---> 22   serialized_pb=_b('\n/tensorflow/core/framework/resource_handle.proto\x12\ntensorflow\"m\n\x0eResourceHandle\x12\x0e\n\x06\x64\x65vice\x18\x01 \x01(\t\x12\x11\n\tcontainer\x18\x02 \x01(\t\x12\x0c\n\x04name\x18\x03 \x01(\t\x12\x11\n\thash_code\x18\x04 \x01(\x04\x12\x17\n\x0fmaybe_type_name\x18\x05 \x01(\tB4\n\x18org.tensorflow.frameworkB\x13ResourceHandleProtoP\x01\xf8\x01\x01\x62\x06proto3')
     23 )
     24 _sym_db.RegisterFileDescriptor(DESCRIPTOR)

TypeError: __init__() got an unexpected keyword argument 'syntax'

I suspect that:
Latest TF version needs protobuf 3.3.0
This DIGITS version runs protobuf-2.6.1

Any advice?
I'm running on Ubuntu 16
Thank you!

Is there any pre-trained models for tensorflow version

I‘m new in deep learning and working on a project for person feature extracting and i think GAN on CelebA Dataset would be a good start point. Is there any pre-trained models for this algorithm and dataset? I have setup a fresh digits env yesterday and start a training job for GAN-CelebA, But I found it's very time consuming(about two weeks) and resources i can apply are very limited. Thank you very much for your help

digits-gan docker image

Hi @gheinrich ,

I am attempting to build a docker file to build in image for digits-gan. I think this would make it a lot easier than installing everything from scratch. I have been trying to familiarize myself with the dockerfile for DIGITS5. Tensorflow requires cuDNN and from what I understand DIGITS5 also requires cuDNN. Since cuDNN is not on aptitude, I am not sure how to properly include it in the dockerfile.

This is what I have added to the DIGITS5 dockerfile so far:

python-h5py \
python-numpy \
python-protobuf \
python-scipy \

and

# DIGITS install
ENV DIGITS_ROOT=~/digits
RUN git clone https://github.com/gheinrich/DIGITS-GAN.git $DIGITS_ROOT
RUN pip install -r $DIGITS_ROOT/requirements.txt

# Tensorflow install
ENV TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-0.12.1-cp27-none-linux_x86_64.whl
RUN pip install --upgrade $TF_BINARY_URL


# Allow plugin and GAN implementation
RUN pip install -e $DIGITS_ROOT
RUN pip install -e $DIGITS_ROOT/plugins/data/gan/
RUN pip install -e $DIGITS_ROOT/plugins/view/gan/

1. when I run docker bin/bash and simply test tensorflow in python interpreter it says it can't find libcudnn and cudnn.
2. when i open the DIGITS webapp and select GAN model there is no tensorflow custom model option.

Do you know what would it take to complete the dockerfile? Also, do you plan to push this "experimental feature" to digits5 soon?

cheers

How to alternate the training process

Hi, and thank you for a great job!
Currently, I'm trying to port DeLiGAN into DIGITS. It has a bit different way of training. The question is where or how can I alternate the training process in DIGITS + TF?
Here's how they train GAN - https://github.com/val-iisc/deligan/blob/master/src/mnist/dg_mnist.py#L180