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

Add DETR (#11653)

Browse Source 
* Squash all commits of modeling_detr_v7 branch into one

* Improve docs

* Fix tests

* Style

* Improve docs some more and fix most tests

* Fix slow tests of ViT, DeiT and DETR

* Improve replacement of batch norm

* Restructure timm backbone forward

* Make DetrForSegmentation support any timm backbone

* Fix name of output

* Address most comments by @LysandreJik

* Give better names for variables

* Conditional imports + timm in setup.py

* Address additional comments by @sgugger

* Make style, add require_timm and require_vision to testsé

* Remove train_backbone attribute of DetrConfig, add methods to freeze/unfreeze backbone

* Add png files to fixtures

* Fix type hint

* Add timm to workflows

* Add `BatchNorm2d` to the weight initialization

* Fix retain_grad test

* Replace model checkpoints by Facebook namespace

* Fix name of checkpoint in test

* Add user-friendly message when scipy is not available

* Address most comments by @patrickvonplaten

* Remove return_intermediate_layers attribute of DetrConfig and simplify Joiner

* Better initialization

* Scipy is necessary to get sklearn metrics

* Rename TimmBackbone to DetrTimmConvEncoder and rename DetrJoiner to DetrConvModel

* Make style

* Improve docs and add 2 community notebooks

Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>
This commit is contained in:
NielsRogge
2021-06-09 17:51:13 +02:00
committed by GitHub
parent d14e0af274
commit d3eacbb829
42 changed files with 5177 additions and 128 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/source/community.md
View File

@@ -59,3 +59,5 @@ This page regroups resources around 🤗 Transformers developed by the community
| [Evaluate LUKE on CoNLL-2003, an important NER benchmark](https://github.com/studio-ousia/luke/blob/master/notebooks/huggingface_conll_2003.ipynb) | How to evaluate *LukeForEntitySpanClassification* on the CoNLL-2003 dataset | [Ikuya Yamada](https://github.com/ikuyamada) |[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/studio-ousia/luke/blob/master/notebooks/huggingface_conll_2003.ipynb) |
| [Evaluate BigBird-Pegasus on PubMed dataset](https://github.com/vasudevgupta7/bigbird/blob/main/notebooks/bigbird_pegasus_evaluation.ipynb) | How to evaluate *BigBirdPegasusForConditionalGeneration* on PubMed dataset | [Vasudev Gupta](https://github.com/vasudevgupta7) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/vasudevgupta7/bigbird/blob/main/notebooks/bigbird_pegasus_evaluation.ipynb) |
| [Speech Emotion Classification with Wav2Vec2](https://github/m3hrdadfi/soxan/blob/main/notebooks/Emotion_recognition_in_Greek_speech_using_Wav2Vec2.ipynb) | How to leverage a pretrained Wav2Vec2 model for Emotion Classification on the MEGA dataset | [Mehrdad Farahani](https://github.com/m3hrdadfi) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/m3hrdadfi/soxan/blob/main/notebooks/Emotion_recognition_in_Greek_speech_using_Wav2Vec2.ipynb) |
| [Detect objects in an image with DETR](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/DETR/DETR_minimal_example_(with_DetrFeatureExtractor).ipynb) | How to use a trained *DetrForObjectDetection* model to detect objects in an image and visualize attention | [Niels Rogge](https://github.com/NielsRogge) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/NielsRogge/Transformers-Tutorials/blob/master/DETR/DETR_minimal_example_(with_DetrFeatureExtractor).ipynb) |
| [Fine-tune DETR on a custom object detection dataset](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/DETR/Fine_tuning_DetrForObjectDetection_on_custom_dataset_(balloon).ipynb) | How to fine-tune *DetrForObjectDetection* on a custom object detection dataset | [Niels Rogge](https://github.com/NielsRogge) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/NielsRogge/Transformers-Tutorials/blob/master/DETR/Fine_tuning_DetrForObjectDetection_on_custom_dataset_(balloon).ipynb) |

88
docs/source/index.rst
View File

@@ -153,128 +153,131 @@ Supported models
19. :doc:`DeiT <model_doc/deit>` (from Facebook) released with the paper `Training data-efficient image transformers &
distillation through attention <https://arxiv.org/abs/2012.12877>`__ by Hugo Touvron, Matthieu Cord, Matthijs
Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
20. :doc:`DialoGPT <model_doc/dialogpt>` (from Microsoft Research) released with the paper `DialoGPT: Large-Scale
20. :doc:`DETR <model_doc/detr>` (from Facebook) released with the paper `End-to-End Object Detection with Transformers
<https://arxiv.org/abs/2005.12872>`__ by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier,
Alexander Kirillov, Sergey Zagoruyko.
21. :doc:`DialoGPT <model_doc/dialogpt>` (from Microsoft Research) released with the paper `DialoGPT: Large-Scale
Generative Pre-training for Conversational Response Generation <https://arxiv.org/abs/1911.00536>`__ by Yizhe
Zhang, Siqi Sun, Michel Galley, Yen-Chun Chen, Chris Brockett, Xiang Gao, Jianfeng Gao, Jingjing Liu, Bill Dolan.
21. :doc:`DistilBERT <model_doc/distilbert>` (from HuggingFace), released together with the paper `DistilBERT, a
22. :doc:`DistilBERT <model_doc/distilbert>` (from HuggingFace), released together with the paper `DistilBERT, a
distilled version of BERT: smaller, faster, cheaper and lighter <https://arxiv.org/abs/1910.01108>`__ by Victor
Sanh, Lysandre Debut and Thomas Wolf. The same method has been applied to compress GPT2 into `DistilGPT2
<https://github.com/huggingface/transformers/tree/master/examples/distillation>`__, RoBERTa into `DistilRoBERTa
<https://github.com/huggingface/transformers/tree/master/examples/distillation>`__, Multilingual BERT into
`DistilmBERT <https://github.com/huggingface/transformers/tree/master/examples/distillation>`__ and a German
version of DistilBERT.
22. :doc:`DPR <model_doc/dpr>` (from Facebook) released with the paper `Dense Passage Retrieval for Open-Domain
23. :doc:`DPR <model_doc/dpr>` (from Facebook) released with the paper `Dense Passage Retrieval for Open-Domain
Question Answering <https://arxiv.org/abs/2004.04906>`__ by Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick
Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih.
23. :doc:`ELECTRA <model_doc/electra>` (from Google Research/Stanford University) released with the paper `ELECTRA:
24. :doc:`ELECTRA <model_doc/electra>` (from Google Research/Stanford University) released with the paper `ELECTRA:
Pre-training text encoders as discriminators rather than generators <https://arxiv.org/abs/2003.10555>`__ by Kevin
Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning.
24. :doc:`FlauBERT <model_doc/flaubert>` (from CNRS) released with the paper `FlauBERT: Unsupervised Language Model
25. :doc:`FlauBERT <model_doc/flaubert>` (from CNRS) released with the paper `FlauBERT: Unsupervised Language Model
Pre-training for French <https://arxiv.org/abs/1912.05372>`__ by Hang Le, Loïc Vial, Jibril Frej, Vincent Segonne,
Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoît Crabbé, Laurent Besacier, Didier Schwab.
25. :doc:`Funnel Transformer <model_doc/funnel>` (from CMU/Google Brain) released with the paper `Funnel-Transformer:
26. :doc:`Funnel Transformer <model_doc/funnel>` (from CMU/Google Brain) released with the paper `Funnel-Transformer:
Filtering out Sequential Redundancy for Efficient Language Processing <https://arxiv.org/abs/2006.03236>`__ by
Zihang Dai, Guokun Lai, Yiming Yang, Quoc V. Le.
26. :doc:`GPT <model_doc/gpt>` (from OpenAI) released with the paper `Improving Language Understanding by Generative
27. :doc:`GPT <model_doc/gpt>` (from OpenAI) released with the paper `Improving Language Understanding by Generative
Pre-Training <https://blog.openai.com/language-unsupervised/>`__ by Alec Radford, Karthik Narasimhan, Tim Salimans
and Ilya Sutskever.
27. :doc:`GPT-2 <model_doc/gpt2>` (from OpenAI) released with the paper `Language Models are Unsupervised Multitask
28. :doc:`GPT-2 <model_doc/gpt2>` (from OpenAI) released with the paper `Language Models are Unsupervised Multitask
Learners <https://blog.openai.com/better-language-models/>`__ by Alec Radford*, Jeffrey Wu*, Rewon Child, David
Luan, Dario Amodei** and Ilya Sutskever**.
28. :doc:`GPT Neo <model_doc/gpt_neo>` (from EleutherAI) released in the repository `EleutherAI/gpt-neo
29. :doc:`GPT Neo <model_doc/gpt_neo>` (from EleutherAI) released in the repository `EleutherAI/gpt-neo
<https://github.com/EleutherAI/gpt-neo>`__ by Sid Black, Stella Biderman, Leo Gao, Phil Wang and Connor Leahy.
29. :doc:`I-BERT <model_doc/ibert>` (from Berkeley) released with the paper `I-BERT: Integer-only BERT Quantization
30. :doc:`I-BERT <model_doc/ibert>` (from Berkeley) released with the paper `I-BERT: Integer-only BERT Quantization
<https://arxiv.org/abs/2101.01321>`__ by Sehoon Kim, Amir Gholami, Zhewei Yao, Michael W. Mahoney, Kurt Keutzer
30. :doc:`LayoutLM <model_doc/layoutlm>` (from Microsoft Research Asia) released with the paper `LayoutLM: Pre-training
31. :doc:`LayoutLM <model_doc/layoutlm>` (from Microsoft Research Asia) released with the paper `LayoutLM: Pre-training
of Text and Layout for Document Image Understanding <https://arxiv.org/abs/1912.13318>`__ by Yiheng Xu, Minghao Li,
Lei Cui, Shaohan Huang, Furu Wei, Ming Zhou.
31. :doc:`LED <model_doc/led>` (from AllenAI) released with the paper `Longformer: The Long-Document Transformer
32. :doc:`LED <model_doc/led>` (from AllenAI) released with the paper `Longformer: The Long-Document Transformer
<https://arxiv.org/abs/2004.05150>`__ by Iz Beltagy, Matthew E. Peters, Arman Cohan.
32. :doc:`Longformer <model_doc/longformer>` (from AllenAI) released with the paper `Longformer: The Long-Document
33. :doc:`Longformer <model_doc/longformer>` (from AllenAI) released with the paper `Longformer: The Long-Document
Transformer <https://arxiv.org/abs/2004.05150>`__ by Iz Beltagy, Matthew E. Peters, Arman Cohan.
33. :doc:`LUKE <model_doc/luke>` (from Studio Ousia) released with the paper `LUKE: Deep Contextualized Entity
34. :doc:`LUKE <model_doc/luke>` (from Studio Ousia) released with the paper `LUKE: Deep Contextualized Entity
Representations with Entity-aware Self-attention <https://arxiv.org/abs/2010.01057>`__ by Ikuya Yamada, Akari Asai,
Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
34. :doc:`LXMERT <model_doc/lxmert>` (from UNC Chapel Hill) released with the paper `LXMERT: Learning Cross-Modality
35. :doc:`LXMERT <model_doc/lxmert>` (from UNC Chapel Hill) released with the paper `LXMERT: Learning Cross-Modality
Encoder Representations from Transformers for Open-Domain Question Answering <https://arxiv.org/abs/1908.07490>`__
by Hao Tan and Mohit Bansal.
35. :doc:`M2M100 <model_doc/m2m_100>` (from Facebook) released with the paper `Beyond English-Centric Multilingual
36. :doc:`M2M100 <model_doc/m2m_100>` (from Facebook) released with the paper `Beyond English-Centric Multilingual
Machine Translation <https://arxiv.org/abs/2010.11125>`__ by by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi
Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman
Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
36. :doc:`MarianMT <model_doc/marian>` Machine translation models trained using `OPUS <http://opus.nlpl.eu/>`__ data by
37. :doc:`MarianMT <model_doc/marian>` Machine translation models trained using `OPUS <http://opus.nlpl.eu/>`__ data by
Jörg Tiedemann. The `Marian Framework <https://marian-nmt.github.io/>`__ is being developed by the Microsoft
Translator Team.
37. :doc:`MBart <model_doc/mbart>` (from Facebook) released with the paper `Multilingual Denoising Pre-training for
38. :doc:`MBart <model_doc/mbart>` (from Facebook) released with the paper `Multilingual Denoising Pre-training for
Neural Machine Translation <https://arxiv.org/abs/2001.08210>`__ by Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li,
Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, Luke Zettlemoyer.
38. :doc:`MBart-50 <model_doc/mbart>` (from Facebook) released with the paper `Multilingual Translation with Extensible
39. :doc:`MBart-50 <model_doc/mbart>` (from Facebook) released with the paper `Multilingual Translation with Extensible
Multilingual Pretraining and Finetuning <https://arxiv.org/abs/2008.00401>`__ by Yuqing Tang, Chau Tran, Xian Li,
Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, Angela Fan.
39. :doc:`Megatron-BERT <model_doc/megatron_bert>` (from NVIDIA) released with the paper `Megatron-LM: Training
40. :doc:`Megatron-BERT <model_doc/megatron_bert>` (from NVIDIA) released with the paper `Megatron-LM: Training
Multi-Billion Parameter Language Models Using Model Parallelism <https://arxiv.org/abs/1909.08053>`__ by Mohammad
Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
40. :doc:`Megatron-GPT2 <model_doc/megatron_gpt2>` (from NVIDIA) released with the paper `Megatron-LM: Training
41. :doc:`Megatron-GPT2 <model_doc/megatron_gpt2>` (from NVIDIA) released with the paper `Megatron-LM: Training
Multi-Billion Parameter Language Models Using Model Parallelism <https://arxiv.org/abs/1909.08053>`__ by Mohammad
Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
41. :doc:`MPNet <model_doc/mpnet>` (from Microsoft Research) released with the paper `MPNet: Masked and Permuted
42. :doc:`MPNet <model_doc/mpnet>` (from Microsoft Research) released with the paper `MPNet: Masked and Permuted
Pre-training for Language Understanding <https://arxiv.org/abs/2004.09297>`__ by Kaitao Song, Xu Tan, Tao Qin,
Jianfeng Lu, Tie-Yan Liu.
42. :doc:`MT5 <model_doc/mt5>` (from Google AI) released with the paper `mT5: A massively multilingual pre-trained
43. :doc:`MT5 <model_doc/mt5>` (from Google AI) released with the paper `mT5: A massively multilingual pre-trained
text-to-text transformer <https://arxiv.org/abs/2010.11934>`__ by Linting Xue, Noah Constant, Adam Roberts, Mihir
Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, Colin Raffel.
43. :doc:`Pegasus <model_doc/pegasus>` (from Google) released with the paper `PEGASUS: Pre-training with Extracted
44. :doc:`Pegasus <model_doc/pegasus>` (from Google) released with the paper `PEGASUS: Pre-training with Extracted
Gap-sentences for Abstractive Summarization <https://arxiv.org/abs/1912.08777>`__> by Jingqing Zhang, Yao Zhao,
Mohammad Saleh and Peter J. Liu.
44. :doc:`ProphetNet <model_doc/prophetnet>` (from Microsoft Research) released with the paper `ProphetNet: Predicting
45. :doc:`ProphetNet <model_doc/prophetnet>` (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.
45. :doc:`Reformer <model_doc/reformer>` (from Google Research) released with the paper `Reformer: The Efficient
46. :doc:`Reformer <model_doc/reformer>` (from Google Research) released with the paper `Reformer: The Efficient
Transformer <https://arxiv.org/abs/2001.04451>`__ by Nikita Kitaev, Łukasz Kaiser, Anselm Levskaya.
46. :doc:`RoBERTa <model_doc/roberta>` (from Facebook), released together with the paper a `Robustly Optimized BERT
47. :doc:`RoBERTa <model_doc/roberta>` (from Facebook), released together with the paper a `Robustly Optimized BERT
Pretraining Approach <https://arxiv.org/abs/1907.11692>`__ by Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar
Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, Veselin Stoyanov.
47. :doc:`RoFormer <model_doc/roformer>` (from ZhuiyiTechnology), released together with the paper a `RoFormer:
48. :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.
48. :doc:`SpeechToTextTransformer <model_doc/speech_to_text>` (from Facebook), released together with the paper
49. :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.
49. :doc:`SqueezeBert <model_doc/squeezebert>` released with the paper `SqueezeBERT: What can computer vision teach NLP
50. :doc:`SqueezeBert <model_doc/squeezebert>` 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.
50. :doc:`T5 <model_doc/t5>` (from Google AI) released with the paper `Exploring the Limits of Transfer Learning with a
51. :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.
51. :doc:`TAPAS <model_doc/tapas>` (from Google AI) released with the paper `TAPAS: Weakly Supervised Table Parsing via
52. :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.
52. :doc:`Transformer-XL <model_doc/transformerxl>` (from Google/CMU) released with the paper `Transformer-XL:
53. :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.
53. :doc:`Vision Transformer (ViT) <model_doc/vit>` (from Google AI) released with the paper `An Image is Worth 16x16
54. :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.
54. :doc:`VisualBERT <model_doc/visual_bert>` (from UCLA NLP) released with the paper `VisualBERT: A Simple and
55. :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.
55. :doc:`Wav2Vec2 <model_doc/wav2vec2>` (from Facebook AI) released with the paper `wav2vec 2.0: A Framework for
56. :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.
56. :doc:`XLM <model_doc/xlm>` (from Facebook) released together with the paper `Cross-lingual Language Model
57. :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.
57. :doc:`XLM-ProphetNet <model_doc/xlmprophetnet>` (from Microsoft Research) released with the paper `ProphetNet:
58. :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.
58. :doc:`XLM-RoBERTa <model_doc/xlmroberta>` (from Facebook AI), released together with the paper `Unsupervised
59. :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.
59. :doc:`XLNet <model_doc/xlnet>` (from Google/CMU) released with the paper `​XLNet: Generalized Autoregressive
60. :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.
60. :doc:`XLSR-Wav2Vec2 <model_doc/xlsr_wav2vec2>` (from Facebook AI) released with the paper `Unsupervised
61. :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.
@@ -318,6 +321,8 @@ Flax), PyTorch, and/or TensorFlow.
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| ConvBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| DETR | ❌ | ❌ | ✅ | ❌ | ❌ |
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| DPR | ✅ | ✅ | ✅ | ✅ | ❌ |
+-----------------------------+----------------+----------------+-----------------+--------------------+--------------+
| DeBERTa | ✅ | ✅ | ✅ | ❌ | ❌ |
@@ -502,6 +507,7 @@ Flax), PyTorch, and/or TensorFlow.
model_doc/deberta
model_doc/deberta_v2
model_doc/deit
model_doc/detr
model_doc/dialogpt
model_doc/distilbert
model_doc/dpr

202
docs/source/model_doc/detr.rst Normal file
View File

@@ -0,0 +1,202 @@
..
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.
DETR
-----------------------------------------------------------------------------------------------------------------------
Overview
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The DETR model was proposed in `End-to-End Object Detection with Transformers <https://arxiv.org/abs/2005.12872>`__ by
Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov and Sergey Zagoruyko. DETR
consists of a convolutional backbone followed by an encoder-decoder Transformer which can be trained end-to-end for
object detection. It greatly simplifies a lot of the complexity of models like Faster-R-CNN and Mask-R-CNN, which use
things like region proposals, non-maximum suppression procedure and anchor generation. Moreover, DETR can also be
naturally extended to perform panoptic segmentation, by simply adding a mask head on top of the decoder outputs.
The abstract from the paper is the following:
*We present a new method that views object detection as a direct set prediction problem. Our approach streamlines the
detection pipeline, effectively removing the need for many hand-designed components like a non-maximum suppression
procedure or anchor generation that explicitly encode our prior knowledge about the task. The main ingredients of the
new framework, called DEtection TRansformer or DETR, are a set-based global loss that forces unique predictions via
bipartite matching, and a transformer encoder-decoder architecture. Given a fixed small set of learned object queries,
DETR reasons about the relations of the objects and the global image context to directly output the final set of
predictions in parallel. The new model is conceptually simple and does not require a specialized library, unlike many
other modern detectors. DETR demonstrates accuracy and run-time performance on par with the well-established and
highly-optimized Faster RCNN baseline on the challenging COCO object detection dataset. Moreover, DETR can be easily
generalized to produce panoptic segmentation in a unified manner. We show that it significantly outperforms competitive
baselines.*
This model was contributed by `nielsr <https://huggingface.co/nielsr>`__. The original code can be found `here
<https://github.com/facebookresearch/detr>`__.
Here's a TLDR explaining how :class:`~transformers.DetrForObjectDetection` works:
First, an image is sent through a pre-trained convolutional backbone (in the paper, the authors use
ResNet-50/ResNet-101). Let's assume we also add a batch dimension. This means that the input to the backbone is a
tensor of shape :obj:`(batch_size, 3, height, width)`, assuming the image has 3 color channels (RGB). The CNN backbone
outputs a new lower-resolution feature map, typically of shape :obj:`(batch_size, 2048, height/32, width/32)`. This is
then projected to match the hidden dimension of the Transformer of DETR, which is :obj:`256` by default, using a
:obj:`nn.Conv2D` layer. So now, we have a tensor of shape :obj:`(batch_size, 256, height/32, width/32).` Next, the
feature map is flattened and transposed to obtain a tensor of shape :obj:`(batch_size, seq_len, d_model)` =
:obj:`(batch_size, width/32*height/32, 256)`. So a difference with NLP models is that the sequence length is actually
longer than usual, but with a smaller :obj:`d_model` (which in NLP is typically 768 or higher).
Next, this is sent through the encoder, outputting :obj:`encoder_hidden_states` of the same shape (you can consider
these as image features). Next, so-called **object queries** are sent through the decoder. This is a tensor of shape
:obj:`(batch_size, num_queries, d_model)`, with :obj:`num_queries` typically set to 100 and initialized with zeros.
These input embeddings are learnt positional encodings that the authors refer to as object queries, and similarly to
the encoder, they are added to the input of each attention layer. Each object query will look for a particular object
in the image. The decoder updates these embeddings through multiple self-attention and encoder-decoder attention layers
to output :obj:`decoder_hidden_states` of the same shape: :obj:`(batch_size, num_queries, d_model)`. Next, two heads
are added on top for object detection: a linear layer for classifying each object query into one of the objects or "no
object", and a MLP to predict bounding boxes for each query.
The model is trained using a **bipartite matching loss**: so what we actually do is compare the predicted classes +
bounding boxes of each of the N = 100 object queries to the ground truth annotations, padded up to the same length N
(so if an image only contains 4 objects, 96 annotations will just have a "no object" as class and "no bounding box" as
bounding box). The `Hungarian matching algorithm <https://en.wikipedia.org/wiki/Hungarian_algorithm>`__ is used to find
an optimal one-to-one mapping of each of the N queries to each of the N annotations. Next, standard cross-entropy (for
the classes) and a linear combination of the L1 and `generalized IoU loss <https://giou.stanford.edu/>`__ (for the
bounding boxes) are used to optimize the parameters of the model.
DETR can be naturally extended to perform panoptic segmentation (which unifies semantic segmentation and instance
segmentation). :class:`~transformers.DetrForSegmentation` adds a segmentation mask head on top of
:class:`~transformers.DetrForObjectDetection`. The mask head can be trained either jointly, or in a two steps process,
where one first trains a :class:`~transformers.DetrForObjectDetection` model to detect bounding boxes around both
"things" (instances) and "stuff" (background things like trees, roads, sky), then freeze all the weights and train only
the mask head for 25 epochs. Experimentally, these two approaches give similar results. Note that predicting boxes is
required for the training to be possible, since the Hungarian matching is computed using distances between boxes.
Tips:
- DETR uses so-called **object queries** to detect objects in an image. The number of queries determines the maximum
number of objects that can be detected in a single image, and is set to 100 by default (see parameter
:obj:`num_queries` of :class:`~transformers.DetrConfig`). Note that it's good to have some slack (in COCO, the
authors used 100, while the maximum number of objects in a COCO image is ~70).
- The decoder of DETR updates the query embeddings in parallel. This is different from language models like GPT-2,
which use autoregressive decoding instead of parallel. Hence, no causal attention mask is used.
- DETR adds position embeddings to the hidden states at each self-attention and cross-attention layer before projecting
to queries and keys. For the position embeddings of the image, one can choose between fixed sinusoidal or learned
absolute position embeddings. By default, the parameter :obj:`position_embedding_type` of
:class:`~transformers.DetrConfig` is set to :obj:`"sine"`.
- During training, the authors of DETR did find it helpful to use auxiliary losses in the decoder, especially to help
the model output the correct number of objects of each class. If you set the parameter :obj:`auxiliary_loss` of
:class:`~transformers.DetrConfig` to :obj:`True`, then prediction feedforward neural networks and Hungarian losses
are added after each decoder layer (with the FFNs sharing parameters).
- If you want to train the model in a distributed environment across multiple nodes, then one should update the
`num_boxes` variable in the `DetrLoss` class of `modeling_detr.py`. When training on multiple nodes, this should be
set to the average number of target boxes across all nodes, as can be seen in the original implementation `here
<https://github.com/facebookresearch/detr/blob/a54b77800eb8e64e3ad0d8237789fcbf2f8350c5/models/detr.py#L227-L232>`__.
- :class:`~transformers.DetrForObjectDetection` and :class:`~transformers.DetrForSegmentation` can be initialized with
any convolutional backbone available in the `timm library <https://github.com/rwightman/pytorch-image-models>`__.
Initializing with a MobileNet backbone for example can be done by setting the :obj:`backbone` attribute of
:class:`~transformers.DetrConfig` to :obj:`"tf_mobilenetv3_small_075"`, and then initializing the model with that
config.
- DETR resizes the input images such that the shortest side is at least a certain amount of pixels while the longest is
at most 1333 pixels. At training time, scale augmentation is used such that the shortest side is randomly set to at
least 480 and at most 800 pixels. At inference time, the shortest side is set to 800. One can use
:class:`~transformers.DetrFeatureExtractor` to prepare images (and optional annotations in COCO format) for the
model. Due to this resizing, images in a batch can have different sizes. DETR solves this by padding images up to the
largest size in a batch, and by creating a pixel mask that indicates which pixels are real/which are padding.
Alternatively, one can also define a custom :obj:`collate_fn` in order to batch images together, using
:meth:`~transformers.DetrFeatureExtractor.pad_and_create_pixel_mask`.
- The size of the images will determine the amount of memory being used, and will thus determine the :obj:`batch_size`.
It is advised to use a batch size of 2 per GPU. See `this Github thread
<https://github.com/facebookresearch/detr/issues/150>`__ for more info.
As a summary, consider the following table:
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
| **Task** | **Object detection** | **Instance segmentation** | **Panoptic segmentation** |
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
| **Description** | Predicting bounding boxes and class labels around | Predicting masks around objects (i.e. instances) in an image | Predicting masks around both objects (i.e. instances) as well as |
| | objects in an image | | "stuff" (i.e. background things like trees and roads) in an image |
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
| **Model** | :class:`~transformers.DetrForObjectDetection` | :class:`~transformers.DetrForSegmentation` | :class:`~transformers.DetrForSegmentation` |
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
| **Example dataset** | COCO detection | COCO detection, | COCO panoptic |
| | | COCO panoptic | |
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
| **Format of annotations to provide to** | {‘image_id’: int, | {‘image_id’: int, | {‘file_name: str, |
| :class:`~transformers.DetrFeatureExtractor` | ‘annotations’: List[Dict]}, each Dict being a COCO | ‘annotations’: [List[Dict]] } (in case of COCO detection) | ‘image_id: int, |
| | object annotation (containing keys "image_id", | | ‘segments_info’: List[Dict] } |
| | | or | |
| | | | and masks_path (path to directory containing PNG files of the masks) |
| | | {‘file_name’: str, | |
| | | ‘image_id’: int, | |
| | | ‘segments_info’: List[Dict]} (in case of COCO panoptic) | |
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
| **Postprocessing** (i.e. converting the | :meth:`~transformers.DetrFeatureExtractor.post_process` | :meth:`~transformers.DetrFeatureExtractor.post_process_segmentation` | :meth:`~transformers.DetrFeatureExtractor.post_process_segmentation`, |
| output of the model to COCO API) | | | :meth:`~transformers.DetrFeatureExtractor.post_process_panoptic` |
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
| **evaluators** | :obj:`CocoEvaluator` with iou_types = “bbox” | :obj:`CocoEvaluator` with iou_types = “bbox”, “segm” | :obj:`CocoEvaluator` with iou_tupes = “bbox, “segm” |
| | | | |
| | | | :obj:`PanopticEvaluator` |
+---------------------------------------------+---------------------------------------------------------+----------------------------------------------------------------------+------------------------------------------------------------------------+
In short, one should prepare the data either in COCO detection or COCO panoptic format, then use
:class:`~transformers.DetrFeatureExtractor` to create :obj:`pixel_values`, :obj:`pixel_mask` and optional
:obj:`labels`, which can then be used to train (or fine-tune) a model. For evaluation, one should first convert the
outputs of the model using one of the postprocessing methods of :class:`~transformers.DetrFeatureExtractor`. These can
be be provided to either :obj:`CocoEvaluator` or :obj:`PanopticEvaluator`, which allow you to calculate metrics like
mean Average Precision (mAP) and Panoptic Quality (PQ). The latter objects are implemented in the `original repository
<https://github.com/facebookresearch/detr>`__. See the example notebooks for more info regarding evaluation.
DETR specific outputs
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.models.detr.modeling_detr.DetrModelOutput
:members:
.. autoclass:: transformers.models.detr.modeling_detr.DetrObjectDetectionOutput
:members:
.. autoclass:: transformers.models.detr.modeling_detr.DetrSegmentationOutput
:members:
DetrConfig
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.DetrConfig
:members:
DetrFeatureExtractor
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.DetrFeatureExtractor
:members: __call__, pad_and_create_pixel_mask, post_process, post_process_segmentation, post_process_panoptic
DetrModel
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.DetrModel
:members: forward
DetrForObjectDetection
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.DetrForObjectDetection
:members: forward
DetrForSegmentation
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. autoclass:: transformers.DetrForSegmentation
:members: forward