Issue 126: IRI-API

.env.example

@@ -1,7 +1,12 @@
+BEAMLINE=8.3.2
 GLOBUS_CLIENT_ID=<globus_client_id>
 GLOBUS_CLIENT_SECRET=<globus_client_secret>
 PREFECT_API_URL=<url_of_prefect_server>
 PREFECT_API_KEY=<prefect_client_secret>
 PUSHGATEWAY_URL=<url_of_pushgateway_server>
 JOB_NAME=<jobname_for_pushgateway>
-INSTANCE_LABEL=<label_for_pushgateway>
+INSTANCE_LABEL=<label_for_pushgateway>
+PATH_NERSC_CLIENT_ID=<path_to_nersc_client_id>
+PATH_NERSC_PRI_KEY=<path_to_nersc_priv_key>
+NERSC_USERNAME=<nersc_username>
+AMSC_API_KEY=<amsc_api_key> # found here: https://profile.american-science-cloud.org/

config.yml

@@ -173,20 +173,24 @@ mlflow:
   staging:
     tracking_uri: https://mlflow-staging.computing.als.lbl.gov
     registry_uri: https://mlflow-staging.computing.als.lbl.gov
+  amsc:
+    tracking_uri: https://mlflow.american-science-cloud.org/
+    registry_uri: https://mlflow.american-science-cloud.org/
+    experiment_name: als-bl832-models
 
 hpc_submission_settings832:
   # ── RECON + MULTIRES SETTINGS ───────────────────────────────────────────────
   nersc_reconstruction:
     # ── SLURM resource allocation ─────────────────────────────────────────────
-    qos: realtime
+    qos: debug
     account: als
-    reservation: "_CAP_TOMO_MOON_CPU"
-    num_nodes: 16
+    reservation: ""
+    num_nodes: 2
     cpus-per-task: 128
     walltime: "0:30:00"
   nersc_multiresolution:
     # ── SLURM resource allocation ─────────────────────────────────────────────
-    qos: realtime
+    qos: debug
     account: als
     reservation: ""
     cpus-per-task: 128
@@ -195,15 +199,15 @@ hpc_submission_settings832:
   # ── PETIOLE SEGMENTATION SETTINGS ───────────────────────────────────────────
   nersc_segmentation_sam3:
     # ── SLURM resource allocation ─────────────────────────────────────────────
-    qos: regular
+    qos: debug
     account: als
     constraint: gpu
     reservation: ""
-    num_nodes: 4
+    num_nodes: 2
     ntasks-per-node: 1
     gpus-per-node: 4
     cpus-per-task: 128
-    walltime: "00:59:00"
+    walltime: "00:30:00"
     # ── Inference parameters ──────────────────────────────────────────────────
     script_name: "src/inference_v6.py"
     batch_size: 1
@@ -226,16 +230,16 @@ hpc_submission_settings832:
     finetuned_checkpoint_path: /global/cfs/cdirs/als/data_mover/8.3.2/tomography_segmentation_scripts/sam3_finetune/sam3/checkpoint_v6.pt
   nersc_segmentation_dinov3:
     # ── SLURM resource allocation ─────────────────────────────────────────────
-    qos: regular
+    qos: debug
     account: als
     constraint: gpu
     reservation: ""
-    num_nodes: 4
+    num_nodes: 2
     ntasks-per-node: 1
     nproc_per_node: 4
     gpus-per-node: 4
     cpus-per-task: 128
-    walltime: "00:59:00"
+    walltime: "00:30:00"
     # ── Inference parameters ──────────────────────────────────────────────────
     script_name: "src.inference_dino_v1"
     batch_size: 4
@@ -246,14 +250,14 @@ hpc_submission_settings832:
     dino_checkpoint_path: /global/cfs/cdirs/als/data_mover/8.3.2/tomography_segmentation_scripts/dino/best.ckpt
   nersc_combine_segmentations:
     # ── SLURM resource allocation ─────────────────────────────────────────────
-    qos: regular
+    qos: debug
     account: als
     constraint: cpu
     reservation: ""
-    num_nodes: 4
+    num_nodes: 2
     ntasks: 128
     cpus-per-task: 1
-    walltime: "01:00:00"
+    walltime: "00:30:00"
     # ── Combination parameters ────────────────────────────────────────────────
     script_name: "src.combine_sam_dino_v3"
     dilate_px: 5