Predeploiment = build and push docker image only

IGNF · Oct 17, 2023 · a7f0323 · a7f0323
1 parent 2808ace
commit a7f0323
Showing 1 changed file with 29 additions and 29 deletions.
diff --git a/.github/workflows/staging-model.yml b/.github/workflows/staging-model.yml
@@ -7,7 +7,7 @@ on:
 
 env:
   nexus_server: 10.128.81.69:8082
-  cicd_path: /var/data/cicd/CICD_github_assets/
+  # cicd_path: /mnt/store-lidarhd/projet-LHD/IA/CICD_github_assets/
   model_id: 20230930_60k_basic_targetted_epoch37_Myria3DV3.4.0
 
 jobs:
@@ -22,35 +22,35 @@ jobs:
     - name: Build docker image
       run: docker build -t myria3d .
 
-    - name: Inference on validation set of 20211001_building_val-lidar-prod-optimization
-      run: >
-        docker run
-        -v ${cicd_path}lidar_prod_optimization/:/lidar_prod_optimization/
-        -v ${cicd_path}20211001_building_val-lidar-prod-optimization/:/data/
-        --ipc=host
-        --shm-size=2gb
-        myria3d
-        python run.py
-        task.task_name=predict
-        predict.src_las=/data/val/*.las
-        predict.output_dir=/lidar_prod_optimization/${model_id}/preds-valset/
-        predict.interpolator.probas_to_save=[building]
-        datamodule.prefetch_factor=1
+    # Do these operations on the GPU-server
 
-    - name: Inference on test set of 20211001_building_val-lidar-prod-optimization
-      run: >
-        docker run
-        -v ${cicd_path}lidar_prod_optimization/:/lidar_prod_optimization/
-        -v ${cicd_path}20211001_building_val-lidar-prod-optimization/:/data/
-        --ipc=host
-        --shm-size=2gb
-        myria3d
-        python run.py
-        task.task_name=predict
-        predict.src_las=/data/test/*.las
-        predict.output_dir=/lidar_prod_optimization/${model_id}/preds-testset/
-        predict.interpolator.probas_to_save=[building]
-        datamodule.prefetch_factor=1
+    # - name: Inference on validation set of 20211001_building_val-lidar-prod-optimization
+    #   run: >
+    #     docker run
+    #     -v ${cicd_path}lidar_prod_optimization/:/lidar_prod_optimization/
+    #     -v ${cicd_path}20211001_building_val-lidar-prod-optimization/:/data/
+    #     --ipc=host
+    #     --shm-size=2gb
+    #     myria3d
+    #     python run.py
+    #     task.task_name=predict
+    #     predict.src_las=/data/val/*.las
+    #     predict.output_dir=/lidar_prod_optimization/${model_id}/preds-valset/
+    #     predict.interpolator.probas_to_save=[building]
+
+    # - name: Inference on test set of 20211001_building_val-lidar-prod-optimization
+    #   run: >
+    #     docker run
+    #     -v ${cicd_path}lidar_prod_optimization/:/lidar_prod_optimization/
+    #     -v ${cicd_path}20211001_building_val-lidar-prod-optimization/:/data/
+    #     --ipc=host
+    #     --shm-size=2gb
+    #     myria3d
+    #     python run.py
+    #     task.task_name=predict
+    #     predict.src_las=/data/test/*.las
+    #     predict.output_dir=/lidar_prod_optimization/${model_id}/preds-testset/
+    #     predict.interpolator.probas_to_save=[building]
 
     # NB: we run the new, tagged image at least once so that is it not prunned
     # by mistake when using docker system prune.