[codex] Move local utils out of git tracking

.gitignore

@@ -131,3 +131,4 @@ job_sub/datasets/**/*_local.*
 # Local-only analysis / figure-generation scripts (notebooks, sensitivity-analysis
 # helpers, etc.). Anything dropped under this folder stays out of git.
 job_sub/utils/analysis/
+utils/

job_sub/conf/config.yaml

@@ -22,15 +22,15 @@ subset_ids_path: ${dataset_field:${datasets_file},${dataset_index},subset_ids_pa
 results_root_dir: /storage2/wangzitongLab/share/deepdraw_opt/jerry
 
 # number of random seeds to run for each active learning experiment
-num_seeds_per_job: 30
+num_seeds_per_job: 20
 seed_start: 0
 parallelize_seeds: true
 
 # active learning settings
 al_settings:
   batch_size: 12
   starting_batch_size: 12
-  max_rounds: 29 # not including the initial selection
+  max_rounds: 19 # not including the initial selection
   feature_transforms: ${feature_transforms}
   target_transforms: ${target_transforms}
   output_dir: ${hydra:runtime.output_dir}
@@ -47,14 +47,14 @@ hydra:
     subdir: ${dataset_name}/${override_values:${hydra.job.override_dirname},dataset_index|single_array_across_datasets|al_settings.seed,default}
   sweeper: # multirun mode sweeps over these parameters
     params:
-      initial_selection_strategy: probcover_euclidean, core_set, random
-      embedding_model: gLM2_166k_kneedle
-      query_strategy: botorch_qlog_nei, topk, botorch_q_ucb
+      initial_selection_strategy: kmedoids
+      embedding_model: 166k_alphagenome_1bp_embeddingkneedle
+      query_strategy: botorch_mes
       predictor: botorch_gp
   launcher:
-    timeout_min: 720
+    timeout_min: 1440
     # everything below is used only for submitting jobs to the cluster
-    partition: intel-sc3,wzt_20250411 # no spacing between names
+    partition: intel-sc3-32c,amd-ep5 # no spacing between names
     cpus_per_task: 1
     qos: huge
     mem_per_cpu: 30GB