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

Add MVP model (#17787)

Browse Source 
* Add MVP model

* Update README

* Remove useless module

* Update docs

* Fix bugs in tokenizer

* Remove useless test

* Remove useless module

* Update vocab

* Remove specifying

* Remove specifying

* Add #Copied ... statement

* Update paper link

* Remove useless TFMvp

* Add #Copied ... statement

* Fix style in test mvp model

* Fix some typos

* Fix properties of unset special tokens in non verbose mode

* Update paper link

* Update MVP doc

* Update MVP doc

* Fix README

* Fix typos in docs

* Update docs
This commit is contained in:
StevenTang1998
2022-06-29 21:30:55 +08:00
committed by GitHub
parent 8f400775fc
commit 3cff4cc587
25 changed files with 4219 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/source/en/_toctree.yml
View File

@@ -308,6 +308,8 @@
title: MPNet
- local: model_doc/mt5
title: MT5
- local: model_doc/mvp
title: MVP
- local: model_doc/nezha
title: NEZHA
- local: model_doc/nystromformer

2
docs/source/en/index.mdx
View File

@@ -124,6 +124,7 @@ The library currently contains JAX, PyTorch and TensorFlow implementations, pret
1. **[MobileBERT](model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
1. **[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.
1. **[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.
1. **[MVP](model_doc/mvp)** (from RUC AI Box) released with the paper [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
1. **[Nezha](model_doc/nezha)** (from Huawei Noah’s Ark Lab) released with the paper [NEZHA: Neural Contextualized Representation for Chinese Language Understanding](https://arxiv.org/abs/1909.00204) by Junqiu Wei, Xiaozhe Ren, Xiaoguang Li, Wenyong Huang, Yi Liao, Yasheng Wang, Jiashu Lin, Xin Jiang, Xiao Chen and Qun Liu.
1. **[Nyströmformer](model_doc/nystromformer)** (from the University of Wisconsin - Madison) released with the paper [Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention](https://arxiv.org/abs/2102.03902) by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, Vikas Singh.
1. **[OPT](master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
@@ -253,6 +254,7 @@ Flax), PyTorch, and/or TensorFlow.
| MobileBERT | ✅ | ✅ | ✅ | ✅ | ❌ |
| MPNet | ✅ | ✅ | ✅ | ✅ | ❌ |
| MT5 | ✅ | ✅ | ✅ | ✅ | ✅ |
| MVP | ✅ | ✅ | ✅ | ❌ | ❌ |
| Nezha | ❌ | ❌ | ✅ | ❌ | ❌ |
| Nyströmformer | ❌ | ❌ | ✅ | ❌ | ❌ |
| OpenAI GPT | ✅ | ✅ | ✅ | ✅ | ❌ |

138
docs/source/en/model_doc/mvp.mdx Normal file
View File

@@ -0,0 +1,138 @@
<!--Copyright 2022 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.
-->
# MVP
## Overview
The MVP model was proposed in [MVP: Multi-task Supervised Pre-training for Natural Language Generation](https://arxiv.org/abs/2206.12131) by Tianyi Tang, Junyi Li, Wayne Xin Zhao and Ji-Rong Wen.
According to the abstract,
- MVP follows a standard Transformer encoder-decoder architecture.
- MVP is supervised pre-trained using labeled datasets.
- MVP also has task-specific soft prompts to stimulate the model's capacity in performing a certain task.
- MVP is specially designed for natural language generation and can be adapted to a wide range of generation tasks, including but not limited to summarization, data-to-text generation, open-ended dialogue system, story generation, question answering, question generation, task-oriented dialogue system, commonsense generation, paraphrase generation, text style transfer, and text simplification. Our model can also be adapted to natural language understanding tasks such as sequence classification and (extractive) question answering.
Tips:
- We have released a series of models [here](https://huggingface.co/models?filter=mvp), including MVP, MVP with task-specific prompts, and multi-task pre-trained variants.
- If you want to use a model without prompts (standard Transformer), you can load it through `MvpForConditionalGeneration.from_pretrained('RUCAIBox/mvp')`.
- If you want to use a model with task-specific prompts, such as summarization, you can load it through `MvpForConditionalGeneration.from_pretrained('RUCAIBox/mvp-summarization')`.
- Our model supports lightweight prompt tuning following [Prefix-tuning](https://arxiv.org/abs/2101.00190) with method `set_lightweight_tuning()`.
This model was contributed by [Tianyi Tang](https://huggingface.co/StevenTang). The detailed information and instructions can be found [here](https://github.com/RUCAIBox/MVP).
## Examples
For summarization, it is an example to use MVP and MVP with summarization-specific prompts.
```python
>>> from transformers import MvpTokenizer, MvpForConditionalGeneration
>>> tokenizer = MvpTokenizer.from_pretrained("RUCAIBox/mvp")
>>> model = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mvp")
>>> model_with_prompt = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mvp-summarization")
>>> inputs = tokenizer(
... "Summarize: You may want to stick it to your boss and leave your job, but don't do it if these are your reasons.",
... return_tensors="pt",
... )
>>> generated_ids = model.generate(**inputs)
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
["Why You Shouldn't Quit Your Job"]
>>> generated_ids = model_with_prompt.generate(**inputs)
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
["Don't do it if these are your reasons"]
```
For data-to-text generation, it is an example to use MVP and multi-task pre-trained variants.
```python
>>> from transformers import MvpTokenizerFast, MvpForConditionalGeneration
>>> tokenizer = MvpTokenizerFast.from_pretrained("RUCAIBox/mvp")
>>> model = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mvp")
>>> model_with_mtl = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mtl-data-to-text")
>>> inputs = tokenizer(
... "Describe the following data: Iron Man | instance of | Superhero [SEP] Stan Lee | creator | Iron Man",
... return_tensors="pt",
... )
>>> generated_ids = model.generate(**inputs)
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
['Stan Lee created the character of Iron Man, a fictional superhero appearing in American comic']
>>> generated_ids = model_with_mtl.generate(**inputs)
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
['Iron Man is a fictional superhero appearing in American comic books published by Marvel Comics.']
```
For lightweight tuning, *i.e.*, fixing the model and only tuning prompts, you can load MVP with randomly initialized prompts or with task-specific prompts. Our code also supports Prefix-tuning with BART following the [original paper](https://arxiv.org/abs/2101.00190).
```python
>>> from transformers import MvpForConditionalGeneration
>>> model = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mvp", use_prompt=True)
>>> # the number of trainable parameters (full tuning)
>>> sum(p.numel() for p in model.parameters() if p.requires_grad)
468116832
>>> # lightweight tuning with randomly initialized prompts
>>> model.set_lightweight_tuning()
>>> # the number of trainable parameters (lightweight tuning)
>>> sum(p.numel() for p in model.parameters() if p.requires_grad)
61823328
>>> # lightweight tuning with task-specific prompts
>>> model = MvpForConditionalGeneration.from_pretrained("RUCAIBox/mtl-data-to-text")
>>> model.set_lightweight_tuning()
>>> # original lightweight Prefix-tuning
>>> model = MvpForConditionalGeneration.from_pretrained("facebook/bart-large", use_prompt=True)
>>> model.set_lightweight_tuning()
```
## MvpConfig
[[autodoc]] MvpConfig
## MvpTokenizer
[[autodoc]] MvpTokenizer
## MvpTokenizerFast
[[autodoc]] MvpTokenizerFast
## MvpModel
[[autodoc]] MvpModel
- forward
## MvpForConditionalGeneration
[[autodoc]] MvpForConditionalGeneration
- forward
## MvpForSequenceClassification
[[autodoc]] MvpForSequenceClassification
- forward
## MvpForQuestionAnswering
[[autodoc]] MvpForQuestionAnswering
- forward
## MvpForCausalLM
[[autodoc]] MvpForCausalLM
- forward

2
docs/source/en/serialization.mdx
View File

@@ -677,4 +677,4 @@ torch.neuron.trace(model, [token_tensor, segments_tensors])
This change enables Neuron SDK to trace the model and optimize it to run in Inf1 instances.
To learn more about AWS Neuron SDK features, tools, example tutorials and latest updates,
please see the [AWS NeuronSDK documentation](https://awsdocs-neuron.readthedocs-hosted.com/en/latest/index.html).
please see the [AWS NeuronSDK documentation](https://awsdocs-neuron.readthedocs-hosted.com/en/latest/index.html).