ai4ia issueshttps://git.embl.de/grp-bio-it-workshops/ai4ia/-/issues2020-10-16T12:24:41Zhttps://git.embl.de/grp-bio-it-workshops/ai4ia/-/issues/5Running CellPose with CPU2020-10-16T12:24:41ZChristian TischerRunning CellPose with CPU@pape
This works :smile_cat:
![image](/uploads/c3786f913c8717f0fa8653f387d5e934/image.png)@pape
This works :smile_cat:
![image](/uploads/c3786f913c8717f0fa8653f387d5e934/image.png)https://git.embl.de/grp-bio-it-workshops/ai4ia/-/issues/4Running CellPose with GPU2020-10-19T15:16:39ZChristian TischerRunning CellPose with GPUcc @pape
I activated the `cellpose-gpu` environment and then within a jupyter notebook I executed these lines of code from the notebook that comes with cellpose: https://github.com/MouseLand/cellpose/blob/master/notebooks/run_cellpose_...cc @pape
I activated the `cellpose-gpu` environment and then within a jupyter notebook I executed these lines of code from the notebook that comes with cellpose: https://github.com/MouseLand/cellpose/blob/master/notebooks/run_cellpose_GPU.ipynb
```python
import numpy as np
import time, os, sys
from urllib.parse import urlparse
import skimage.io
import matplotlib.pyplot as plt
import matplotlib as mpl
%matplotlib inline
mpl.rcParams['figure.dpi'] = 300
from urllib.parse import urlparse
import mxnet as mx
from cellpose import utils
use_GPU = utils.use_gpu()
print('GPU activated? %d'%use_GPU)
```
But I got as a response:
```
CUDA version not installed/working, will use CPU version.
GPU activated? 0
```https://git.embl.de/grp-bio-it-workshops/ai4ia/-/issues/3Jupyter disk quota2020-10-19T09:20:38ZChristian TischerJupyter disk quotaStarting Jupyter with `jupyter notebook --ip $IP` can yield below error:
```
(cellpose-gpu) tischer@gpu16:/g/cba/tischer/cellpose/notebooks$ jupyter notebook --ip $IP
[I 14:09:39.681 NotebookApp] Serving notebooks from local directory: ...Starting Jupyter with `jupyter notebook --ip $IP` can yield below error:
```
(cellpose-gpu) tischer@gpu16:/g/cba/tischer/cellpose/notebooks$ jupyter notebook --ip $IP
[I 14:09:39.681 NotebookApp] Serving notebooks from local directory: /g/cba/tischer/cellpose/notebooks
[I 14:09:39.681 NotebookApp] Jupyter Notebook 6.1.4 is running at:
[I 14:09:39.681 NotebookApp] http://10.11.12.76:8888/?token=40972faaf891cd7f4718e6a5e94fb7a4d692c56b9e935fee
[I 14:09:39.681 NotebookApp] or http://127.0.0.1:8888/?token=40972faaf891cd7f4718e6a5e94fb7a4d692c56b9e935fee
[I 14:09:39.681 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[E 14:09:39.728 NotebookApp] Failed to write server-info to /home/tischer/.local/share/jupyter/runtime/nbserver-70802.json: [Errno 122] Disk quota exceeded
```
I think we should find out how to use another directory for the jupyter tmp files, because our home folders at EMBL are very small.
cc @papehttps://git.embl.de/grp-bio-it-workshops/ai4ia/-/issues/2Running CellPose2020-10-19T09:21:50ZChristian TischerRunning CellPose@pape
Regrading running CellPose:
Does every user need to run this?
```
conda env create -f environment_gpu.yaml
```
Couldn't we persistently set up the two environments for them?
- `cellpose-gpu`
- `cellpose-cpu`
For stardist we do...@pape
Regrading running CellPose:
Does every user need to run this?
```
conda env create -f environment_gpu.yaml
```
Couldn't we persistently set up the two environments for them?
- `cellpose-gpu`
- `cellpose-cpu`
For stardist we do provide the environment, don't we?https://git.embl.de/grp-bio-it-workshops/ai4ia/-/issues/1Launching jupyter notebook from interactive cluster job2020-07-14T10:40:55ZChristian TischerLaunching jupyter notebook from interactive cluster job@pape @heriche @pecar
I am starting this issue to document/ discuss attempts launching jupyter notebooks as interactive GPU cluster jobs.
I talked with Jure and he pointed me to below documentation:
- https://wiki.embl.de/cluster/Env...@pape @heriche @pecar
I am starting this issue to document/ discuss attempts launching jupyter notebooks as interactive GPU cluster jobs.
I talked with Jure and he pointed me to below documentation:
- https://wiki.embl.de/cluster/Env#Graphical_Access
- https://wiki.embl.de/cluster/Slurm#Interactive_jobs
My plan is to try to use above documentation and run:
- `source /g/almf/software/miniconda3/run_conda.sh`
- `conda activate stardist-gpu`
- `cd /g/cba/exchange/stardist/examples/2D`
- `jupyter notebook 2_training.ipynb`
I'll let you know how it goes, but I am also happy to get any kind of input.