Skip to content

Latest commit

 

History

History
57 lines (41 loc) · 2.27 KB

commands.md

File metadata and controls

57 lines (41 loc) · 2.27 KB

Tensorflow

Installation

Installing Tensorflow 0.12.1 (Python 2.7, Ubuntu 64, with GPU)

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

Installing Tensorflow 0.12.1 (Python 3.4, Ubuntu 64, with GPU)

export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/gpu/tensorflow_gpu-0.12.1-cp34-cp34m-linux_x86_64.whl
pip3 install --upgrade $TF_BINARY_URL

Tensorboad

Running tensorboard without docker:

python3 /usr/local/lib/python3.4/dist-packages/tensorflow/tensorboard/tensorboard.py --logdir=/data/rothfuss/training/

Accessing Tensorboard on other pc via ssh

ssh -L 6007:127.0.0.1:6006 rothfuss@xxxxx032

virtualenv

Run python 3.4 virtual environment:

source ~/p3.4/bin/activate source /localhome/rothfuss/p3.4_local

Using Docker

Container content: TensorFlow r0.12.0 rc1 CUDA8.0 cuDNN 5 Python 3.4.3

Run command:

nvidia-docker run --rm -ti ferreirafabio/deepepisodicmemory

Run command with ~/Downloads directory of host OS attached

nvidia-docker run -v ~Downloads:/Downloads --rm -ti ferreirafabio/deepepisodicmemory

Run command with local dir attached

nvidia-docker run -v /localhome/rothfuss:/local --rm -ti ferreirafabio/deepepisodicmemory

Then run training:

python3 train_model.py --path /local/data/ArtificialFlyingBlobs/tfrecords --output_dir /local/training --num_epochs 80000

Load trained model:

python3 train_model.py --path /local/data/ArtificialFlyingBlobs/tfrecords --output_dir /local/training --num_epochs 80000 --pretrained_model /local/training/

Running tensorboard:

tensorboard --logdir=/local/training/log/ --port 6006 python3 /usr/local/lib/python3.4/dist-packages/tensorflow/tensorboard/tensorboard.py --logdir=/data/rothfuss/training

Explicit command to origin:

nvidia-docker run -it -p 8888:8888 tensorflow/tensorflow:0.12.0-rc1-devel-gpu-py3

newer version:

nvidia-docker run -it tensorflow/tensorflow:1.0.0-devel-gpu-py3

save a docker image and commit it to docker hub:
  1. run image, make changes to the image
  2. in another terminal, run "nvidia-docker ps -a"
  3. get the container_id
  4. run "nvidia-docker commit <container_id> ferreirafabio/deepepisodicmemory