1. wandb project name

2. sdp vocabs auto build
3. fix srl distill

ltp/__init__.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*_
 # Author: Yunlong Feng <[email protected]>
 
-__version__ = '4.1.3.post1'
+__version__ = '4.1.4'
 
 from . import const
 from . import nn, utils

ltp/data/dataset/conllu.py

@@ -40,7 +40,9 @@ def build_vocabs(data_dir, train_file, dev_file=None, test_file=None, min_freq=5
         'word': (1, Counter()), 'lemma': (2, Counter()), 'upos': (3, Counter()),
         'xpos': (4, Counter()), 'feats': (5, Counter()), 'deprel': (7, Counter()),
         # FOR CHAR FEATS
-        'word_char': (1, Counter())
+        'word_char': (1, Counter()),
+        # DEPS
+        'deps': (8, Counter())
     }
 
     if any([os.path.exists(os.path.join(data_dir, 'vocabs', f'{key}.txt')) for key in counters]):
@@ -56,6 +58,9 @@ def build_vocabs(data_dir, train_file, dev_file=None, test_file=None, min_freq=5
             for name, (row, counter) in counters.items():
                 if 'char' in name:
                     counter.update(itertools.chain(*values[row]))
+                elif 'deps' == name:
+                    deps = [[label.split(':', maxsplit=1)[1] for label in dep.split('|')] for dep in values[row]]
+                    counter.update(itertools.chain(*deps))
                 else:
                     counter.update(values[row])
 
@@ -95,12 +100,17 @@ def _info(self):
             'xpos': self.config.xpos,
             'feats': self.config.feats,
             'deprel': self.config.deprel,
+            'deps': self.config.deps
         }
 
         for key in feats:
             if feats[key] is None:
                 feats[key] = os.path.join(self.config.data_dir, 'vocabs', f'{key}.txt')
 
+        deps_rel_feature = create_feature(feats['deps'])
+        if deps_rel_feature.num_classes > 1:
+            self.config.deps = feats['deps']
+
         return datasets.DatasetInfo(
             description=_DESCRIPTION,
             features=datasets.Features(
@@ -117,8 +127,8 @@ def _info(self):
                         {
                             'id': datasets.Value('int64'),
                             'head': datasets.Value("int64"),
-                            'rel': create_feature(self.config.deps)
-                        } if self.config.deps else create_feature(self.config.deps)
+                            'rel': deps_rel_feature
+                        } if deps_rel_feature.num_classes > 1 else create_feature(None)
                     ),
                     "misc": datasets.Sequence(datasets.Value("string")),
                 }

ltp/multitask.py

@@ -200,15 +200,18 @@ def build_ner_distill_dataset(args):
     model.eval()
     model.freeze()
 
-    dataset, metric = task_named_entity_recognition.build_dataset(model, args.ner_data_dir, task_info.task_name)
+    dataset, metric = task_named_entity_recognition.build_dataset(
+        model, args.ner_data_dir,
+        task_named_entity_recognition.task_info.task_name
+    )
     train_dataloader = torch.utils.data.DataLoader(
         dataset[datasets.Split.TRAIN],
         batch_size=args.batch_size,
         collate_fn=collate,
         num_workers=args.num_workers
     )
 
-    output = os.path.join(args.ner_data_dir, task_info.task_name, 'output.npz')
+    output = os.path.join(args.ner_data_dir, task_named_entity_recognition.task_info.task_name, 'output.npz')
 
     if torch.cuda.is_available():
         model.cuda()
@@ -246,6 +249,7 @@ def add_task_specific_args(parent_parser):
     parser.add_argument('--seed', type=int, default=19980524)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
     parser.add_argument('--patience', type=int, default=5)
     parser.add_argument('--batch_size', type=int, default=8)
     parser.add_argument('--gpus_per_trial', type=float, default=1.0)

ltp/multitask_distill.py

@@ -90,16 +90,11 @@ def distill_linear(batch, result, target, temperature_scheduler, model: Model =
         transitions = torch.as_tensor(extra['transitions'], device=model.device)
         end_transitions = torch.as_tensor(extra['end_transitions'], device=model.device)
 
-        temperature = temperature_scheduler(active_logits, active_target_logits)
-        kd_loss = kd_mse_loss(active_logits, active_target_logits, temperature)
-
-        transitions_temp = temperature_scheduler(model.srl_classifier.crf.transitions, transitions)
-        s_transitions_temp = temperature_scheduler(model.srl_classifier.crf.start_transitions, start_transitions)
-        e_transitions_temp = temperature_scheduler(model.srl_classifier.crf.end_transitions, end_transitions)
+        kd_loss = kd_mse_loss(active_logits, active_target_logits)
 
-        crf_loss = kd_mse_loss(transitions, model.srl_classifier.crf.transitions, transitions_temp) + \
-                   kd_mse_loss(start_transitions, model.srl_classifier.crf.start_transitions, s_transitions_temp) + \
-                   kd_mse_loss(end_transitions, model.srl_classifier.crf.end_transitions, e_transitions_temp)
+        crf_loss = kd_mse_loss(transitions, model.srl_classifier.crf.transitions) + \
+                   kd_mse_loss(start_transitions, model.srl_classifier.crf.start_transitions) + \
+                   kd_mse_loss(end_transitions, model.srl_classifier.crf.end_transitions)
 
         return kd_loss + crf_loss
     else:
@@ -168,7 +163,7 @@ def distill_matrix_crf(batch, result, target, temperature_scheduler, model: Mode
     index = logits_mask[:, 0]
     logits_mask = logits_mask[index]
 
-    s_rel, labels = result.arc_logits, result.labels
+    s_rel, labels = result.rel_logits, result.labels
     t_rel = target
 
     active_logits = s_rel[logits_mask]
@@ -181,13 +176,13 @@ def distill_matrix_crf(batch, result, target, temperature_scheduler, model: Mode
     transitions = torch.as_tensor(extra['transitions'], device=model.device)
     end_transitions = torch.as_tensor(extra['end_transitions'], device=model.device)
 
-    transitions_temp = temperature_scheduler(model.srl_classifier.crf.transitions, transitions)
-    s_transitions_temp = temperature_scheduler(model.srl_classifier.crf.start_transitions, start_transitions)
-    e_transitions_temp = temperature_scheduler(model.srl_classifier.crf.end_transitions, end_transitions)
+    # transitions_temp = temperature_scheduler(model.srl_classifier.crf.transitions, transitions)
+    # s_transitions_temp = temperature_scheduler(model.srl_classifier.crf.start_transitions, start_transitions)
+    # e_transitions_temp = temperature_scheduler(model.srl_classifier.crf.end_transitions, end_transitions)
 
-    crf_loss = kd_mse_loss(transitions, model.srl_classifier.crf.transitions, transitions_temp) + \
-               kd_mse_loss(start_transitions, model.srl_classifier.crf.start_transitions, s_transitions_temp) + \
-               kd_mse_loss(end_transitions, model.srl_classifier.crf.end_transitions, e_transitions_temp)
+    crf_loss = kd_mse_loss(transitions, model.srl_classifier.crf.transitions) + \
+               kd_mse_loss(start_transitions, model.srl_classifier.crf.start_transitions) + \
+               kd_mse_loss(end_transitions, model.srl_classifier.crf.end_transitions)
     return kd_loss + crf_loss
 
 
@@ -359,6 +354,7 @@ def add_task_specific_args(parent_parser):
     parser.add_argument('--seed', type=int, default=19980524)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
 
     parser.add_argument('--disable_seg', action='store_true')
     parser.add_argument('--disable_pos', action='store_true')

ltp/task_dependency_parsing.py

@@ -141,6 +141,7 @@ def add_task_specific_args(parent_parser):
     parser = ArgumentParser(parents=[parent_parser], add_help=False)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
     parser.add_argument('--patience', type=int, default=5)
     parser.add_argument('--seed', type=int, default=19980524)
     parser.add_argument('--gpus_per_trial', type=float, default=1.0)

ltp/task_named_entity_recognition.py

@@ -129,6 +129,7 @@ def add_task_specific_args(parent_parser):
     parser = ArgumentParser(parents=[parent_parser], add_help=False)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
     parser.add_argument('--patience', type=int, default=5)
     parser.add_argument('--seed', type=int, default=19980524)
     parser.add_argument('--gpus_per_trial', type=float, default=1.0)

ltp/task_part_of_speech.py

@@ -117,6 +117,7 @@ def add_task_specific_args(parent_parser):
     parser = ArgumentParser(parents=[parent_parser], add_help=False)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
     parser.add_argument('--patience', type=int, default=5)
     parser.add_argument('--gpus_per_trial', type=float, default=1.0)
     parser.add_argument('--cpus_per_trial', type=float, default=1.0)

ltp/task_segmention.py

@@ -119,6 +119,7 @@ def add_task_specific_args(parent_parser):
     parser = ArgumentParser(parents=[parent_parser], add_help=False)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
     parser.add_argument('--seed', type=int, default=19980524)
     parser.add_argument('--patience', type=int, default=5)
     parser.add_argument('--gpus_per_trial', type=float, default=1.0)

ltp/task_semantic_dependency_parsing.py

@@ -162,6 +162,7 @@ def add_task_specific_args(parent_parser):
     parser = ArgumentParser(parents=[parent_parser], add_help=False)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
     parser.add_argument('--seed', type=int, default=19980524)
     parser.add_argument('--patience', type=int, default=5)
     parser.add_argument('--gpus_per_trial', type=float, default=1.0)

ltp/task_semantic_role_labeling.py

@@ -134,6 +134,7 @@ def add_task_specific_args(parent_parser):
     parser = ArgumentParser(parents=[parent_parser], add_help=False)
     parser.add_argument('--tune', action='store_true')
     parser.add_argument('--offline', action='store_true')
+    parser.add_argument('--project', type=str, default='ltp')
     parser.add_argument('--patience', type=int, default=5)
     parser.add_argument('--gpus_per_trial', type=float, default=1.0)
     parser.add_argument('--cpus_per_trial', type=float, default=1.0)

ltp/utils/common_train.py

@@ -43,7 +43,7 @@ def common_train(args, model_class, task_info: TaskInfo, build_method=default_bu
     try:
         import wandb
         logger = loggers.WandbLogger(
-            project='ltp4',
+            project=args.project,
             save_dir='lightning_logs',
             name=f'{task_info.task_name}_{this_time}',
             offline=args.offline