Use parallel experiments with configurable worker

nemo_skills/pipeline/utils/exp.py

@@ -57,13 +57,21 @@ def get_exp_handles(expname: str, ignore_finished=True, ignore_exp_not_exists=Tr
           is called, but finish before nemo-run submits a new job (which might take minutes)
     """
 
-    def _get_handles(exp):
+    def _get_handles(exp: run.Experiment):
         handles = []
-        for job in exp.jobs:
+        status_dict = exp.status(return_dict=True)
+        assert status_dict, f"No status found for experiment {exp._id}"
+        for _, status_info in status_dict.items():
             if not ignore_finished or (
-                job.status(exp._runner) in [AppState.RUNNING, AppState.PENDING, AppState.SUBMITTED, AppState.UNKNOWN]
+                status_info['status']
+                in [
+                    AppState.RUNNING,
+                    AppState.PENDING,
+                    AppState.SUBMITTED,
+                    AppState.UNKNOWN,
+                ]
             ):
-                handles.append(job.handle)
+                handles.append(status_info['handle'])
                 continue
         return handles
 
@@ -611,7 +619,12 @@ def get_exp(expname, cluster_config, _reuse_exp=None):
     # nemo-run redefines the handlers, so removing ours to avoid duplicate logs
     remove_handlers()
     if cluster_config['executor'] == 'slurm':
-        return run.Experiment(expname)
+        return run.Experiment(
+            expname,
+            skip_status_at_exit=True,
+            serialize_metadata_for_scripts=False,
+            threadpool_workers=cluster_config.get('num_workers', 4),
+        )
     # hiding all nemo-run logs otherwise as they are not useful locally
     if cluster_config['executor'] == 'local':
         return run.Experiment(expname, clean_mode=True)