returnn 1.20250922.155846__py3-none-any.whl → 1.20251006.114241__py3-none-any.whl

returnn/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: returnn
-Version: 1.20250922.155846
+Version: 1.20251006.114241
 Summary: The RWTH extensible training framework for universal recurrent neural networks
 Home-page: https://github.com/rwth-i6/returnn/
 Author: Albert Zeyer

returnn/_setup_info_generated.py

@@ -1,2 +1,2 @@
-version = '1.20250922.155846'
-long_version = '1.20250922.155846+git.2334ec6'
+version = '1.20251006.114241'
+long_version = '1.20251006.114241+git.7745ba7'

returnn/datasets/generating.py

@@ -7,14 +7,13 @@ from __future__ import annotations
 from typing import Optional, Union, Any, Sequence, List, Tuple, Dict
 import numpy
 import sys
-import typing
 
 from returnn.util.basic import class_idx_seq_to_1_of_k, CollectionReadCheckCovered
 from returnn.log import log
 from returnn.tensor import Tensor, Dim, TensorDict
 
 from .util.feature_extraction import ExtractAudioFeatures
-from .util.vocabulary import *
+from .util.vocabulary import Vocabulary, BytePairEncoding, CharacterTargets
 from .audio import OggZipDataset  # noqa # for API compatibility
 from .basic import Dataset, DatasetSeq, convert_data_dims
 from .cached2 import CachedDataset2
@@ -1280,12 +1279,6 @@ class CopyTaskDataset(GeneratingDataset):
         return DatasetSeq(seq_idx=seq_idx, features=seq_np, targets={"classes": seq_np})
 
 
-# Multiple external sources where we could write automatic wrappers:
-# * https://github.com/tensorflow/datasets
-# * tf.contrib.keras.datasets, https://www.tensorflow.org/api_docs/python/tf/keras/datasets
-# * nltk.corpus
-
-
 class TimitDataset(CachedDataset2):
     """
     DARPA TIMIT Acoustic-Phonetic Continuous Speech Corpus.
@@ -1553,7 +1546,7 @@ class TimitDataset(CachedDataset2):
 
         self._random_permute_audio = CollectionReadCheckCovered.from_bool_or_dict(random_permute_audio)
 
-        self._seq_order = None  # type: typing.Optional[typing.Sequence[int]]
+        self._seq_order: Optional[Sequence[int]] = None
         self._init_timit()
 
         self._audio_data = {}  # seq_tag -> (audio, sample_rate). loaded by self._reader_thread_main
@@ -1927,8 +1920,8 @@ class BlissDataset(CachedDataset2):
         self._with_delta = with_delta
         self.num_inputs *= 1 + with_delta
         self._bpe_file = open(bpe_file, "r")
-        self._seqs = []  # type: typing.List[BlissDataset.SeqInfo]
-        self._vocab = {}  # type: typing.Dict[str,int]  # set in self._parse_vocab
+        self._seqs: List[BlissDataset.SeqInfo] = []
+        self._vocab: Dict[str, int] = {}  # set in self._parse_vocab
         self._parse_bliss_xml(filename=path)
         # TODO: loading audio like in TimitDataset, and in parallel
         self._bpe = BytePairEncoding(vocab_file=vocab_file, bpe_file=bpe_file)
@@ -2100,7 +2093,7 @@ class LibriSpeechCorpus(CachedDataset2):
             self.targets = CharacterTargets(**chars)
         elif targets is None:
             assert bpe is None and chars is None
-            self.targets = None  # type: typing.Optional[Vocabulary]
+            self.targets: Optional[Vocabulary] = None
         else:
             raise Exception("invalid targets %r. provide bpe or chars" % targets)
         if self.targets:
@@ -2128,7 +2121,7 @@ class LibriSpeechCorpus(CachedDataset2):
             self._reference_seq_order = seqs
             self.transs = {s: self.transs[s] for s in seqs}
         self.epoch_wise_filter = epoch_wise_filter
-        self._seq_order = None  # type: typing.Optional[typing.Sequence[int]]
+        self._seq_order: Optional[Sequence[int]] = None
         self.init_seq_order()
 
     def _collect_trans(self):
@@ -2294,9 +2287,9 @@ class LibriSpeechCorpus(CachedDataset2):
         """:return: whether this dataset supports sharding"""
         return True
 
-    def get_current_seq_order(self):
+    def get_current_seq_order(self) -> Sequence[int]:
         """
-        :rtype: typing.Sequence[int]
+        :return: seq order of current epoch
         """
         assert self._seq_order is not None
         return self._seq_order
@@ -2446,7 +2439,7 @@ class Enwik8Corpus(CachedDataset2):
         self._batch_num_seqs = batch_num_seqs
         self._random = numpy.random.RandomState(1)  # seed will be set in init_seq_order
         self._seq_starts = numpy.arange(0, len(self._data) - 1, seq_len)
-        self._seq_order = None  # type: typing.Optional[typing.Sequence[int]]
+        self._seq_order: Optional[Sequence[int]] = None
 
     def get_data_dtype(self, key):
         """

{returnn-1.20250922.155846.dist-info → returnn-1.20251006.114241.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: returnn
-Version: 1.20250922.155846
+Version: 1.20251006.114241
 Summary: The RWTH extensible training framework for universal recurrent neural networks
 Home-page: https://github.com/rwth-i6/returnn/
 Author: Albert Zeyer

{returnn-1.20250922.155846.dist-info → returnn-1.20251006.114241.dist-info}/RECORD

@@ -1,9 +1,9 @@
-returnn/PKG-INFO,sha256=_tLrRBsotctlK2JO9venoFFVIKDm_CaT1s_KtSjmCf0,5215
+returnn/PKG-INFO,sha256=khjeqHtAYC68kPEh0ltnHLB14E2_UOObt3MGLvgeMTY,5215
 returnn/__init__.py,sha256=biBtRsM0WZ406vShaeH-9WFoqJ8XwTbn6g0EeFJ7l8E,1012
 returnn/__main__.py,sha256=lHyZcu_0yc9f7Vf_Kfdy9PmeU0T76XVXnpalHi5WKro,31740
 returnn/__old_mod_loader__.py,sha256=nvsNY-xELdS_IPNkv66Q9Rmvg4dbGW0-EBRDcCmctos,7654
 returnn/__setup__.py,sha256=22kQn2fh11iPM0hLb2Fy5sLmoU1JGvmDxXRYuRgQkwU,4659
-returnn/_setup_info_generated.py,sha256=doToEY43Ijf3bmda-mfSLIIrU58QQanqyQOyZQfp8hw,77
+returnn/_setup_info_generated.py,sha256=MFwUI_cce-3opoV_eKfpRd891cRwBo6RMbTGriz-HTg,77
 returnn/config.py,sha256=3tmKhB6FnQZaNdtcYsiB61JnEY--iZ2qmJ4yq0b6tE0,29140
 returnn/forward_iface.py,sha256=A_OJiaXsX4MlXQRzST86ylyxSUZbC402PQL1REcqHjM,911
 returnn/learning_rate_control.py,sha256=ZvWryAn_tv9DhV8sh1LV3eE34Yltl3On3mYZAG4hR9s,34684
@@ -18,7 +18,7 @@ returnn/datasets/bundle_file.py,sha256=KQNrS1MSf-4_idlK0c0KFwON-f5sEK0sWU15WpoMY
 returnn/datasets/cached.py,sha256=RyefRjSDdp-HveK-2vLy2C6BIHcpqQ_lNvUKlIa4QAI,25412
 returnn/datasets/cached2.py,sha256=oJOq2lWRQpxm6kyUKW1w5qZBd4kdKEpwM7KY_QnXbq4,11922
 returnn/datasets/distrib_files.py,sha256=-WNVhtvdJFP3L9Meh33oTSYc0FJSvF40mJ5UI_vJbSE,30233
-returnn/datasets/generating.py,sha256=9U_w6URIrv-Rb-hDbPOzYW9qYXzJbw32N6G268IKyoM,99833
+returnn/datasets/generating.py,sha256=Qb7V94N_GfL2pZPxWS5PmzszoVXXKzuUmsHuW3dmVbc,99556
 returnn/datasets/hdf.py,sha256=v5sjBenURR9Z-g7AQ9tsL84yDSye5RtbLpym3M6HSDE,67833
 returnn/datasets/lm.py,sha256=rQ3jV43lSnlGkKu7m5jTTH7aK0BOMXQocsHfJ8OGec8,99950
 returnn/datasets/map.py,sha256=kOBJVZmwDhLsOplzDNByIfa0NRSUaMo2Lsy36lBvxrM,10907
@@ -253,8 +253,8 @@ returnn/util/sig_proc.py,sha256=Tjz0VOAVyqu2qDCF5HZ1JjALjcFsHcNkcd96WgZeKfE,7265
 returnn/util/task_system.py,sha256=y4sMVXQ25Qd2z0rx03uOlXlkE-jbCYC1Sjfn-XlraVU,26003
 returnn/util/train_proc_manager.py,sha256=Pjht28k6uz6BNQ47uW6Gf880iyq5q4wx7P_K2tmoAM8,3266
 returnn/util/watch_memory.py,sha256=BR5P2kvBN6UI81cE0_1WAA6Hd1SByLbBaiDxvLhPOew,4213
-returnn-1.20250922.155846.dist-info/LICENSE,sha256=ywBD_U2aD4vpuoIgNAsjIGBYydl0tVKll3De0Z8s77c,11041
-returnn-1.20250922.155846.dist-info/METADATA,sha256=_tLrRBsotctlK2JO9venoFFVIKDm_CaT1s_KtSjmCf0,5215
-returnn-1.20250922.155846.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
-returnn-1.20250922.155846.dist-info/top_level.txt,sha256=Lsn4WZc5Pbfk0-xDQOgnFCxOoqxL4CyeM3N1TFbJncw,8
-returnn-1.20250922.155846.dist-info/RECORD,,
+returnn-1.20251006.114241.dist-info/LICENSE,sha256=ywBD_U2aD4vpuoIgNAsjIGBYydl0tVKll3De0Z8s77c,11041
+returnn-1.20251006.114241.dist-info/METADATA,sha256=khjeqHtAYC68kPEh0ltnHLB14E2_UOObt3MGLvgeMTY,5215
+returnn-1.20251006.114241.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
+returnn-1.20251006.114241.dist-info/top_level.txt,sha256=Lsn4WZc5Pbfk0-xDQOgnFCxOoqxL4CyeM3N1TFbJncw,8
+returnn-1.20251006.114241.dist-info/RECORD,,