ryuji0123 / cv_dnn Goto Github PK
View Code? Open in Web Editor NEWLicense: MIT License
License: MIT License
cloneしたのち
./build_docker_image.sh
を実行すると実行途中で以下のようなエラーが起こります。
Step 17/19 : RUN $SETTING_PATH/setup.sh
---> Running in 4c2d0d3af719
/bin/sh: 1: /duser/settings/setup.sh: not found
The command '/bin/sh -c $SETTING_PATH/setup.sh' returned a non-zero code: 127
とりあえずhttps://github.com/ryuji0123/settings をcloneして解決しましたが,ReadMeにこの手順を追加するか,settingsなしでも実行できるようにして欲しいです。
add optuna with pytorch lightning
None
None
add pytest to test functions
None
None
I am trying to run
./run_docker_container.sh
but it will have an error message:
Error response from daemon: No such container: cv_dnn
"--gpus" requires API version 1.40, but the Docker daemon API version is 1.39
The whole message is as follows:
$ ./build_docker_image.sh
Sending build context to Docker daemon 398.3kB
Step 1/19 : FROM pytorch/pytorch
---> 349148663741
Step 2/19 : ENV DEBIAN_FRONTEND noninteractive
---> Using cache
---> 606755e73492
Step 3/19 : RUN apt-get update -qq && apt-get install -y curl libopencv-dev lsof git sudo tmux tree vim wget zsh && apt-get clean && rm -rf /var/lib/apt/lists/* && rm -rf /var/cache/apk/*
---> Using cache
---> 18cc2cdb7ed4
Step 4/19 : ENV WORK_PATH /workspace
---> Using cache
---> 94e70d8d5333
Step 5/19 : WORKDIR $WORK_PATH
---> Using cache
---> 0b02b62b8735
Step 6/19 : ENV PYTHONPATH "/workspace:${PYTHONPATH}"
---> Using cache
---> bd3ecf7fd142
Step 7/19 : ENV PIP_OPTIONS "--no-cache-dir --progress-bar off"
---> Using cache
---> 3c8b9432d78f
Step 8/19 : COPY docker/requirements.txt $WORK_PATH/docker/
---> Using cache
---> 2955e134c831
Step 9/19 : RUN pip install ${PIP_OPTIONS} -r $WORK_PATH/docker/requirements.txt && pip install ${PIP_OPTIONS} -U setuptools
---> Using cache
---> a3f06e21b091
Step 10/19 : ARG USER_ID
---> Using cache
---> a60c0c4f3a1b
Step 11/19 : ARG GROUP_ID
---> Using cache
---> 96845eff52c7
Step 12/19 : RUN addgroup --gid $GROUP_ID duser && adduser --disabled-password --gecos '' --uid $USER_ID --gid $GROUP_ID duser && adduser duser sudo && echo '%sudo ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers
---> Using cache
---> 2f0ebe2869b0
Step 13/19 : USER duser
---> Using cache
---> 24eea61e4120
Step 14/19 : ENV SETTING_PATH /duser/settings
---> Using cache
---> c436624f00e5
Step 15/19 : COPY docker/settings $SETTING_PATH
---> 672bd8def242
Step 16/19 : WORKDIR $SETTING_PATH
---> Running in 168161460adb
Removing intermediate container 168161460adb
---> 639901b9ea1f
Step 17/19 : RUN $SETTING_PATH/setup.sh
---> Running in 0b87df537d2a
Install to "/home/duser/.vim/dein/repos/github.com/Shougo/dein.vim"...
git is /usr/bin/git
Begin fetching dein...
Cloning into '/home/duser/.vim/dein/repos/github.com/Shougo/dein.vim'...
Done.
Please add the following settings for dein to the top of your vimrc (Vim) or init.vim (NeoVim) file:
"dein Scripts-----------------------------
if &compatible
set nocompatible " Be iMproved
endif
" Required:
set runtimepath+=/home/duser/.vim/dein/repos/github.com/Shougo/dein.vim
" Required:
if dein#load_state('/home/duser/.vim/dein')
call dein#begin('/home/duser/.vim/dein')
" Let dein manage dein
" Required:
call dein#add('/home/duser/.vim/dein/repos/github.com/Shougo/dein.vim')
" Add or remove your plugins here like this:
"call dein#add('Shougo/neosnippet.vim')
"call dein#add('Shougo/neosnippet-snippets')
" Required:
call dein#end()
call dein#save_state()
endif
" Required:
filetype plugin indent on
syntax enable
" If you want to install not installed plugins on startup.
"if dein#check_install()
" call dein#install()
"endif
"End dein Scripts-------------------------
Done.
Complete setup dein!
Removing intermediate container 0b87df537d2a
---> f8a8db256c81
Step 18/19 : SHELL ["/bin/zsh", "-c"]
---> Running in 56fb03db74b3
Removing intermediate container 56fb03db74b3
---> d43c9c421bec
Step 19/19 : WORKDIR $WORK_PATH
---> Running in 63c954568f67
Removing intermediate container 63c954568f67
---> f29ebf535e82
Successfully built f29ebf535e82
Successfully tagged cv_dnn:latest
ryoto@gorgon:~/cv_dnn$ ./run_docker_container.sh
Error response from daemon: No such container: cv_dnn
"--gpus" requires API version 1.40, but the Docker daemon API version is 1.39
ryoto@gorgon:~/cv_dnn$
My GPU is
GPU: Tesla K80, CUDA10.0, cuDNNv7.5.0
add flake8 to check coding style
None
None
add codes to visualize mean & variance of logged metrics such as train_loss
None
Make process of mlflow starts at the same time the container starts to run.
tomoino/PyTorch-Project-Template: docker/run.sh
tomoino/PyTorch-Project-Template: docker/init.sh
add codes to visualize continuous error bands of loss curve and other metrics while training
None
None
update README
None
None
Add template files and write some instructions on how to use them.
None
None
Replace yacs with hydra. You can see skew_mixup. Basically, you can copy and paste them. Also, add multiple datasets and models in skew_mixup such as cifar100 and ResNet18.
(train.py)
trainer = pl.Trainer(
callbacks=[checkpoint_callback],
distributed_backend=args.TRAIN.DISTRIBUTED_BACKEND, # unnecessary
gpus=args.TRAIN.GPUS, # change to 1
logger=mlflow_logger,
max_epochs=args.TRAIN.MAX_EPOCHS,
replace_sampler_ddp=False, # unnecessary
)
(config/defaults.py)
_C.TRAIN.DISTRIBUTED_BACKEND = 'ddp' # unnecessary
_C.TRAIN.GPUS = 2 # unnecessary
#!/bin/sh
TIMESTAMP=`date +%Y-%m-%d_%H-%M-%S`
TMP_RESULTS_DIR="$(pwd)/.tmp_results/${TIMESTAMP}"
TRAIN_LOG_FILE="${TMP_RESULTS_DIR}/log.txt"
mkdir -p $TMP_RESULTS_DIR
export CUDA_VISIBLE_DEVICES=0
nohup python -u train.py \
-m data=cifar10,cifar100 model=alexnet,resnet18 \
> $TRAIN_LOG_FILE &
sleep 1s
tail -f $TRAIN_LOG_FILE
finally:
run_id = mlflow_logger.run_id
if run_id is not None:
with open(args_file_path, 'w') as f: # unnecessary
with redirect_stdout(f):
print(args.dump())
mlflow_client = MlflowClient()
mlflow_client.log_artifact(run_id, args_file_path) # unnecessary
mlflow_client.log_artifact(run_id, train_log_file_path) # unnecessary
if exist_error:
mlflow_client.log_artifact(run_id, error_file_path)
rmtree(tmp_results_dir, ignore_errors=True) # maybe unnecessary
Of course, if you know how to save logs and config files generated by hydra in mlflow and delete them, it's nice to do it in this week.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.