Open outofstyle opened 3 years ago
@outofstyle
`./nas/cars/cars.yml' is here: https://github.com/huawei-noah/vega/blob/master/examples/nas/cars/cars.yml
The quick start documents is out of date. We'll fix it soon. Please comment out the following line:
lr_scheduler:
type: MultiStepLR
params:
# warmup: False
milestones: [30]
gamma: 0.5
run the following command
vega ./nas/cars/cars.yml
Whycommand not found: vega
?
Thx!
@sisrfeng
Generally, the command is installed in the ~/.local/bin directory. If the --user parameter is used during the installation. pip adds this directory to $PATH during installation. You can try to login out and then login in. Or add ~/.local/bin to $PATH.
Is there a binary file called vega? But I can not find it
And why can't I install the lateset version(1.7)?
~/.local/bin is already in $PATH But still command not found: vega
@sisrfeng
If you use the following command, vega will existed in /home/wf/.local/bin
.
pip install --user --upgrade noah-vega
If you not use --user
parameter, vega will existed in /data/wf/anaconda3/envs/py38_torch18/lib/python3.8/bin
.
pip install --upgrade noah-vega
Both of these directories are in your $PATH variable.
You can use python -m vega.tools.run_pipeline
instead of vega
Vega follows the general python package rules and does not perform any special operations. Packaging script: https://github.com/huawei-noah/vega/blob/master/setup.py I don't know why vega command is not in environment.
How to get the latest version ?
I use:
pip install --upgrade noah-vega
but get version 1.2:
which requires:
Where is your pypi repo? Maybe your pypi repo did not sync with pypi original repo.
You can download the last version(noah_vega-1.7.1-py3-none-any.whl) from release page: https://github.com/huawei-noah/vega/releases
The execute the following command:
pip install ./noah_vega-1.7.1-py3-none-any.whl
I noticed that your environment is python 3.8 and torch 1.8. The vega automatically installs the torch1.3 during installation. This is not what you want, so it is recommended not to install the dependency:
pip install --no-deps ./noah_vega-1.7.1-py3-none-any.whl
Then install the dependency separately.
pip install click
pip install distributed
pip install numpy
pip install opencv-python-headless
pip install pandas
pip install pareto
pip install pillow
pip install psutil
pip install py-dag
pip install PyYAML
pip install pyzmq
pip install scikit-learn
pip install scipy
pip install tensorboardX
pip install thop
The next version of Vega does not specify the version number of the torch.
Under python3.7 and 3.6, I can install vega normally by pip install --upgrade noah-vega
But under 3.8 and 3.9, that command installs vega of version1.2 or 1.0, and no binary vega
can be found.
@sisrfeng
Using this command:
pip install --no-deps ./noah_vega-1.7.1-py3-none-any.whl
What's the error messages?
@sisrfeng
Using this command:
pip install --no-deps ./noah_vega-1.7.1-py3-none-any.whl
What's the error messages?
No errror for this command. I am just curious why :
Under python3.7 and 3.6, I can install vega normally by
pip install --upgrade noah-vega
But under 3.8 and 3.9, that command installs vega of version1.2 or 1.0, and no binaryvega
can be found.
@sisrfeng
But under 3.8 and 3.9, that command installs vega of version1.2 or 1.0, -- I have no idea. I add "--no-deps" param when installing vega.
no binary vega can be found. -- Vega 1.0 or Vega 1.2 have no binary command "vega".
Maybe pip install git+
is more convinient (and do the same thing as in the screen shot?):
pip install git+https://github.com/huawei-noah/vega.git
Im a new user of vega ,but i cannot understand where to run ./nas/cars/cars.yml. And for example simpleCnn in quick start, i create it on nas according to examples. But ,it has a error that TypeError: init() got an unexpected keyword argument 'warmup'.