Skip to content
4 changes: 2 additions & 2 deletions mlproject/MLproject
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ entry_points:
default: /p/scratch/deepacf/maelstrom/maelstrom_data/ap1/air_temperature
datasize:
type: string
default: 5GB
default: 5TB
mode:
type: string
default: train
Expand All @@ -26,7 +26,7 @@ entry_points:
default: 3
command: >
python3 -u benchmark.py
{datadir}/{datasize}/202???01T*.nc
{datadir}/{datasize}/20200301T*.nc
-m {mode}
-b {batch_size}
-p {patch_size}
Expand Down
5 changes: 2 additions & 3 deletions mlproject/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -25,10 +25,9 @@ pip install -r maelstrom-train/benchmark/requirements_wo_modules.txt

5. The results will be logged to an Experiment on the MLflow tracking server on Mantik. Set up a project in Mantik and create a new Experiment. Note its experiment Id, which will be needed in the submission command. For a step-by-step guide, refer to the Quickstart tutorial available [here](https://mantik-ai.gitlab.io/mantik/ui/quickstart.html).

6. Update the `unicore-config-venv.yaml` file by specifying the `PreRunCommand` with the path to your virtual environment.
6. Update the `unicore-config-venv.yaml` file by specifying the `PreRunCommandOnComputeNode` with the path to your virtual environment.

<pre><code> PreRunCommand:
Command: >
<pre><code> PreRunCommandOnComputeNode: >
module load Stages/2022 GCCcore/.11.2.0 NCCL/2.11.4-CUDA-11.5 Python/3.9.6;
source <b>/path/to/&lt;venv-name&gt;</b>/bin/activate;
</code></pre>
Expand Down
25 changes: 11 additions & 14 deletions mlproject/unicore-config-venv.yaml
Original file line number Diff line number Diff line change
@@ -1,18 +1,15 @@
UnicoreApiUrl: https://zam2125.zam.kfa-juelich.de:9112/JUWELS/rest/core
Exclude:
- '*.sif'
- figures
- saved_model
- README.md
Resources:
Nodes: 1
Queue: develbooster
Environment:
PreRunCommand:
Command: >
module load Stages/2022 GCCcore/.11.2.0 NCCL/2.11.4-CUDA-11.5 Python/3.9.6;
source /p/project/deepacf/maelstrom/grau1/ap1/bin/activate;
ExecuteOnLoginNode: False
Variables:
GIT_PYTHON_REFRESH: quiet
Resources:
Queue: devel
Nodes: 1
Exclude:
- "*.sif"
- "figures"
- "saved_model"
- README.md

PreRunCommandOnComputeNode: >
module load Stages/2022 GCCcore/.11.2.0 NCCL/2.11.4-CUDA-11.5 Python/3.9.6;
source /p/scratch/deepacf/maelstrom/grau1/ap1_env/bin/activate