SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Add the SEW and SEW-D speech models (#13962)

Browse Source 
* Working encoder

* SEW-D and tests

* Further conv fixes

* Automodels and conv inits

* Update integration tests, add docs

* Docs cleanup, resolve todos

* Conf fix

* Fix docs

* Fix tests, apply suggestions

* Update src/transformers/models/sew/modeling_sew.py

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Model conversion and updated no-mask tests

* Remove copy of feature_proj

* Style

* Update src/transformers/models/auto/feature_extraction_auto.py

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Update src/transformers/models/auto/feature_extraction_auto.py

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Move orgs

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
This commit is contained in:
Anton Lozhkov
2021-10-15 18:26:26 +03:00
committed by GitHub
parent d5b82bb70c
commit cd3166a8ed
24 changed files with 4991 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
docs/source/index.rst
View File

@@ -268,59 +268,65 @@ Supported models
57. :doc:`RoFormer <model_doc/roformer>` (from ZhuiyiTechnology), released together with the paper a `RoFormer:
Enhanced Transformer with Rotary Position Embedding <https://arxiv.org/pdf/2104.09864v1.pdf>`__ by Jianlin Su and
Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
58. :doc:`SpeechToTextTransformer <model_doc/speech_to_text>` (from Facebook), released together with the paper
58. :doc:`SEW <model_doc/sew>` (from ASAPP) released with the paper `Performance-Efficiency Trade-offs in Unsupervised
Pre-training for Speech Recognition <https://arxiv.org/abs/2109.06870>`__ by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu
Han, Kilian Q. Weinberger, Yoav Artzi.
59. :doc:`SEW-D <model_doc/sew_d>` (from ASAPP) released with the paper `Performance-Efficiency Trade-offs in
Unsupervised Pre-training for Speech Recognition <https://arxiv.org/abs/2109.06870>`__ by Felix Wu, Kwangyoun Kim,
Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
60. :doc:`SpeechToTextTransformer <model_doc/speech_to_text>` (from Facebook), released together with the paper
`fairseq S2T: Fast Speech-to-Text Modeling with fairseq <https://arxiv.org/abs/2010.05171>`__ by Changhan Wang, Yun
Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, Juan Pino.
59. :doc:`SpeechToTextTransformer2 <model_doc/speech_to_text_2>` (from Facebook), released together with the paper
61. :doc:`SpeechToTextTransformer2 <model_doc/speech_to_text_2>` (from Facebook), released together with the paper
`Large-Scale Self- and Semi-Supervised Learning for Speech Translation <https://arxiv.org/abs/2104.06678>`__ by
Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau.
60. :doc:`Splinter <model_doc/splinter>` (from Tel Aviv University), released together with the paper `Few-Shot
62. :doc:`Splinter <model_doc/splinter>` (from Tel Aviv University), released together with the paper `Few-Shot
Question Answering by Pretraining Span Selection <https://arxiv.org/abs/2101.00438>`__ by Ori Ram, Yuval Kirstain,
Jonathan Berant, Amir Globerson, Omer Levy.
61. :doc:`SqueezeBert <model_doc/squeezebert>` (from Berkeley) released with the paper `SqueezeBERT: What can computer
63. :doc:`SqueezeBert <model_doc/squeezebert>` (from Berkeley) released with the paper `SqueezeBERT: What can computer
vision teach NLP about efficient neural networks? <https://arxiv.org/abs/2006.11316>`__ by Forrest N. Iandola,
Albert E. Shaw, Ravi Krishna, and Kurt W. Keutzer.
62. :doc:`T5 <model_doc/t5>` (from Google AI) released with the paper `Exploring the Limits of Transfer Learning with a
64. :doc:`T5 <model_doc/t5>` (from Google AI) released with the paper `Exploring the Limits of Transfer Learning with a
Unified Text-to-Text Transformer <https://arxiv.org/abs/1910.10683>`__ by Colin Raffel and Noam Shazeer and Adam
Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi Zhou and Wei Li and Peter J. Liu.
63. :doc:`T5v1.1 <model_doc/t5v1.1>` (from Google AI) released in the repository
65. :doc:`T5v1.1 <model_doc/t5v1.1>` (from Google AI) released in the repository
`google-research/text-to-text-transfer-transformer
<https://github.com/google-research/text-to-text-transfer-transformer/blob/main/released_checkpoints.md#t511>`__ by
Colin Raffel and Noam Shazeer and Adam Roberts and Katherine Lee and Sharan Narang and Michael Matena and Yanqi
Zhou and Wei Li and Peter J. Liu.
64. :doc:`TAPAS <model_doc/tapas>` (from Google AI) released with the paper `TAPAS: Weakly Supervised Table Parsing via
66. :doc:`TAPAS <model_doc/tapas>` (from Google AI) released with the paper `TAPAS: Weakly Supervised Table Parsing via
Pre-training <https://arxiv.org/abs/2004.02349>`__ by Jonathan Herzig, Paweł Krzysztof Nowak, Thomas Müller,
Francesco Piccinno and Julian Martin Eisenschlos.
65. :doc:`Transformer-XL <model_doc/transformerxl>` (from Google/CMU) released with the paper `Transformer-XL:
67. :doc:`Transformer-XL <model_doc/transformerxl>` (from Google/CMU) released with the paper `Transformer-XL:
Attentive Language Models Beyond a Fixed-Length Context <https://arxiv.org/abs/1901.02860>`__ by Zihang Dai*,
Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
66. `TrOCR <https://huggingface.co/transformers/master/model_doc/trocr.html>`__ (from Microsoft), released together
68. `TrOCR <https://huggingface.co/transformers/master/model_doc/trocr.html>`__ (from Microsoft), released together
with the paper `TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models
<https://arxiv.org/abs/2109.10282>`__ by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang,
Zhoujun Li, Furu Wei.
67. :doc:`Vision Transformer (ViT) <model_doc/vit>` (from Google AI) released with the paper `An Image is Worth 16x16
69. :doc:`Vision Transformer (ViT) <model_doc/vit>` (from Google AI) released with the paper `An Image is Worth 16x16
Words: Transformers for Image Recognition at Scale <https://arxiv.org/abs/2010.11929>`__ by Alexey Dosovitskiy,
Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias
Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, Neil Houlsby.
68. :doc:`VisualBERT <model_doc/visual_bert>` (from UCLA NLP) released with the paper `VisualBERT: A Simple and
70. :doc:`VisualBERT <model_doc/visual_bert>` (from UCLA NLP) released with the paper `VisualBERT: A Simple and
Performant Baseline for Vision and Language <https://arxiv.org/pdf/1908.03557>`__ by Liunian Harold Li, Mark
Yatskar, Da Yin, Cho-Jui Hsieh, Kai-Wei Chang.
69. :doc:`Wav2Vec2 <model_doc/wav2vec2>` (from Facebook AI) released with the paper `wav2vec 2.0: A Framework for
71. :doc:`Wav2Vec2 <model_doc/wav2vec2>` (from Facebook AI) released with the paper `wav2vec 2.0: A Framework for
Self-Supervised Learning of Speech Representations <https://arxiv.org/abs/2006.11477>`__ by Alexei Baevski, Henry
Zhou, Abdelrahman Mohamed, Michael Auli.
70. :doc:`XLM <model_doc/xlm>` (from Facebook) released together with the paper `Cross-lingual Language Model
72. :doc:`XLM <model_doc/xlm>` (from Facebook) released together with the paper `Cross-lingual Language Model
Pretraining <https://arxiv.org/abs/1901.07291>`__ by Guillaume Lample and Alexis Conneau.
71. :doc:`XLM-ProphetNet <model_doc/xlmprophetnet>` (from Microsoft Research) released with the paper `ProphetNet:
73. :doc:`XLM-ProphetNet <model_doc/xlmprophetnet>` (from Microsoft Research) released with the paper `ProphetNet:
Predicting Future N-gram for Sequence-to-Sequence Pre-training <https://arxiv.org/abs/2001.04063>`__ by Yu Yan,
Weizhen Qi, Yeyun Gong, Dayiheng Liu, Nan Duan, Jiusheng Chen, Ruofei Zhang and Ming Zhou.
72. :doc:`XLM-RoBERTa <model_doc/xlmroberta>` (from Facebook AI), released together with the paper `Unsupervised
74. :doc:`XLM-RoBERTa <model_doc/xlmroberta>` (from Facebook AI), released together with the paper `Unsupervised
Cross-lingual Representation Learning at Scale <https://arxiv.org/abs/1911.02116>`__ by Alexis Conneau*, Kartikay
Khandelwal*, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke
Zettlemoyer and Veselin Stoyanov.
73. :doc:`XLNet <model_doc/xlnet>` (from Google/CMU) released with the paper `​XLNet: Generalized Autoregressive
75. :doc:`XLNet <model_doc/xlnet>` (from Google/CMU) released with the paper `​XLNet: Generalized Autoregressive
Pretraining for Language Understanding <https://arxiv.org/abs/1906.08237>`__ by Zhilin Yang*, Zihang Dai*, Yiming
Yang, Jaime Carbonell, Ruslan Salakhutdinov, Quoc V. Le.
74. :doc:`XLSR-Wav2Vec2 <model_doc/xlsr_wav2vec2>` (from Facebook AI) released with the paper `Unsupervised
76. :doc:`XLSR-Wav2Vec2 <model_doc/xlsr_wav2vec2>` (from Facebook AI) released with the paper `Unsupervised
Cross-Lingual Representation Learning For Speech Recognition <https://arxiv.org/abs/2006.13979>`__ by Alexis
Conneau, Alexei Baevski, Ronan Collobert, Abdelrahman Mohamed, Michael Auli.
@@ -446,6 +452,10 @@ Flax), PyTorch, and/or TensorFlow.
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| RoFormer | ✅ | ✅ | ✅ | ✅ | ❌ |
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| SEW | ❌ | ❌ | ✅ | ❌ | ❌ |
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| SEW-D | ❌ | ❌ | ✅ | ❌ | ❌ |
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| Speech Encoder decoder | ❌ | ❌ | ✅ | ❌ | ❌ |
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| Speech2Text | ✅ | ❌ | ✅ | ❌ | ❌ |
@@ -621,6 +631,8 @@ Flax), PyTorch, and/or TensorFlow.
model_doc/retribert
model_doc/roberta
model_doc/roformer
model_doc/sew
model_doc/sew_d
model_doc/speechencoderdecoder
model_doc/speech_to_text
model_doc/speech_to_text_2

61
docs/source/model_doc/sew.rst Normal file
View File

@@ -0,0 +1,61 @@
..
Copyright 2021 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
SEW
-----------------------------------------------------------------------------------------------------------------------
Overview
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
SEW (Squeezed and Efficient Wav2Vec) was proposed in `Performance-Efficiency Trade-offs in Unsupervised Pre-training
for Speech Recognition <https://arxiv.org/abs/2109.06870>`__ by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q.
Weinberger, Yoav Artzi.
The abstract from the paper is the following:
*This paper is a study of performance-efficiency trade-offs in pre-trained models for automatic speech recognition
(ASR). We focus on wav2vec 2.0, and formalize several architecture designs that influence both the model performance
and its efficiency. Putting together all our observations, we introduce SEW (Squeezed and Efficient Wav2vec), a
pre-trained model architecture with significant improvements along both performance and efficiency dimensions across a
variety of training setups. For example, under the 100h-960h semi-supervised setup on LibriSpeech, SEW achieves a 1.9x
inference speedup compared to wav2vec 2.0, with a 13.5% relative reduction in word error rate. With a similar inference
time, SEW reduces word error rate by 25-50% across different model sizes.*
Tips:
- SEW is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
- SEWForCTC is fine-tuned using connectionist temporal classification (CTC) so the model output has to be decoded using
:class:`~transformers.Wav2Vec2CTCTokenizer`.
This model was contributed by `anton-l <https://huggingface.co/anton-l>`__.
SEWConfig
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.SEWConfig
:members:
SEWModel
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.SEWModel
:members: forward
SEWForCTC
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.SEWForCTC
:members: forward

61
docs/source/model_doc/sew_d.rst Normal file
View File

@@ -0,0 +1,61 @@
..
Copyright 2021 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
SEW-D
-----------------------------------------------------------------------------------------------------------------------
Overview
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
SEW-D (Squeezed and Efficient Wav2Vec with Disentangled attention) was proposed in `Performance-Efficiency Trade-offs
in Unsupervised Pre-training for Speech Recognition <https://arxiv.org/abs/2109.06870>`__ by Felix Wu, Kwangyoun Kim,
Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
The abstract from the paper is the following:
*This paper is a study of performance-efficiency trade-offs in pre-trained models for automatic speech recognition
(ASR). We focus on wav2vec 2.0, and formalize several architecture designs that influence both the model performance
and its efficiency. Putting together all our observations, we introduce SEW (Squeezed and Efficient Wav2vec), a
pre-trained model architecture with significant improvements along both performance and efficiency dimensions across a
variety of training setups. For example, under the 100h-960h semi-supervised setup on LibriSpeech, SEW achieves a 1.9x
inference speedup compared to wav2vec 2.0, with a 13.5% relative reduction in word error rate. With a similar inference
time, SEW reduces word error rate by 25-50% across different model sizes.*
Tips:
- SEW-D is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
- SEWDForCTC is fine-tuned using connectionist temporal classification (CTC) so the model output has to be decoded
using :class:`~transformers.Wav2Vec2CTCTokenizer`.
This model was contributed by `anton-l <https://huggingface.co/anton-l>`__.
SEWDConfig
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.SEWDConfig
:members:
SEWDModel
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.SEWDModel
:members: forward
SEWDForCTC
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.SEWDForCTC
:members: forward