Use fine-tuning schedule to train NER models

ahisto_named_entity_search/default.ini

@@ -102,3 +102,6 @@ schedule = fair-sequential-schedule
 
 [recognition.FairSequentialSchedule]
 maximum_number_of_training_epochs_per_objective = 1
+
+[recognition.FineTuningSchedule]
+maximum_number_of_training_epochs_per_objective = 5

ahisto_named_entity_search/recognition/schedule.py

@@ -30,6 +30,25 @@ class FairSequentialSchedule(SequentialSchedule):
                     yield objective
 
 
+class FineTuningSchedule(SequentialSchedule):
+    CONFIG = _CONFIG['recognition.FineTuningSchedule']
+    MAX_NUM_TRAIN_EPOCHS = CONFIG.getint('maximum_number_of_training_epochs_per_objective')
+
+    label = 'fine_tuning'
+
+    def _sample_objectives(self, split: str) -> Iterable[Objective]:
+        for objective in self.objectives[split].values():
+            starting_epoch = objective.epoch
+            while True:
+                if objective in self.converged_objectives and not self.args.log_converged_objectives:
+                    break
+                if split == 'train':
+                    num_train_epochs = objective.epoch - starting_epoch
+                    if num_train_epochs >= self.MAX_NUM_TRAIN_EPOCHS:
+                        break
+                yield objective
+
+
 def get_schedule(schedule_name: str, objectives: Iterable[Objective],
                  adaptation_arguments: AdaptationArguments) -> Schedule:
     objectives = list(objectives)
@@ -37,6 +56,8 @@ def get_schedule(schedule_name: str, objectives: Iterable[Objective],
         schedule = SequentialSchedule(objectives, adaptation_arguments)
     elif schedule_name == 'fair-sequential':
         schedule = FairSequentialSchedule(objectives, adaptation_arguments)
+    elif schedule_name == 'fine-tuning':
+        schedule = FineTuningSchedule(objectives, adaptation_arguments)
     elif schedule_name == 'parallel':
         schedule = ParallelSchedule(objectives, adaptation_arguments)
     else:

scripts/03_train_ner_models.sh

@@ -8,7 +8,7 @@ IMAGE_NAME=ahisto/named-entity-search:latest
 ROOT_PATH=/nlp/projekty/ahisto/public_html/named-entity-search/results/
 ANNOTATION_PATH=/nlp/projekty/ahisto/annotations/
 OCR_EVAL_PATH=/nlp/projekty/ahisto/ahisto-ocr-eval
-SCHEDULE_NAME=parallel
+SCHEDULE_NAME=fine-tuning
 
 DOCKER_BUILDKIT=1 docker build --build-arg UID="$(id -u)" --build-arg GID="$(id -g)" --build-arg UNAME="$(id -u -n)" . -f scripts//03_train_ner_models.Dockerfile -t "$IMAGE_NAME"