Add DePlot + MatCha on transformers (#22528)

* add deplot + matcha on `transformers`

* more docs

* correct path

* Update docs/source/en/model_doc/deplot.mdx

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* fix

* use auto processor

* Update docs/source/en/model_doc/matcha.mdx

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* make fixup

* Update docs/source/en/model_doc/deplot.mdx

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* add correct names

---------

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

docs/source/en/_toctree.yml

@@ -572,6 +572,8 @@
         title: CLIPSeg
       - local: model_doc/data2vec
         title: Data2Vec
+      - local: model_doc/deplot
+        title: DePlot
       - local: model_doc/donut
         title: Donut
       - local: model_doc/flava
@@ -592,6 +594,8 @@
         title: LiLT
       - local: model_doc/lxmert
         title: LXMERT
+      - local: model_doc/matcha
+        title: MatCha
       - local: model_doc/mgp-str
         title: MGP-STR
       - local: model_doc/oneformer

docs/source/en/index.mdx

@@ -92,6 +92,7 @@ The documentation is organized into five sections:
 1. **[Decision Transformer](model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
 1. **[Deformable DETR](model_doc/deformable_detr)** (from SenseTime Research) released with the paper [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://arxiv.org/abs/2010.04159) by Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, Jifeng Dai.
 1. **[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.
+1. **[DePlot](model_doc/deplot)** (from Google AI) released with the paper [DePlot: One-shot visual language reasoning by plot-to-table translation](https://arxiv.org/abs/2212.10505) by Fangyu Liu, Julian Martin Eisenschlos, Francesco Piccinno, Syrine Krichene, Chenxi Pang, Kenton Lee, Mandar Joshi, Wenhu Chen, Nigel Collier, Yasemin Altun.
 1. **[DETA](model_doc/deta)** (from The University of Texas at Austin) released with the paper [NMS Strikes Back](https://arxiv.org/abs/2212.06137) by Jeffrey Ouyang-Zhang, Jang Hyun Cho, Xingyi Zhou, Philipp Krähenbühl.
 1. **[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.
 1. **[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.
@@ -149,6 +150,7 @@ The documentation is organized into five sections:
 1. **[MarkupLM](model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
 1. **[Mask2Former](model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.
 1. **[MaskFormer](model_doc/maskformer)** (from Meta and UIUC) released with the paper [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://arxiv.org/abs/2107.06278) by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov.
+1. **[MatCha](model_doc/matcha)** (from Google AI) released with the paper [MatCha: Enhancing Visual Language Pretraining with Math Reasoning and Chart Derendering](https://arxiv.org/abs/2212.09662) by Fangyu Liu, Francesco Piccinno, Syrine Krichene, Chenxi Pang, Kenton Lee, Mandar Joshi, Yasemin Altun, Nigel Collier, Julian Martin Eisenschlos.
 1. **[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.
 1. **[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.
 1. **[MEGA](model_doc/mega)** (from Facebook) released with the paper [Mega: Moving Average Equipped Gated Attention](https://arxiv.org/abs/2209.10655) by Xuezhe Ma, Chunting Zhou, Xiang Kong, Junxian He, Liangke Gui, Graham Neubig, Jonathan May, and Luke Zettlemoyer.

docs/source/en/model_doc/deplot.mdx

@@ -0,0 +1,58 @@
+<!--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.
+-->
+
+# DePlot
+
+## Overview 
+
+DePlot was proposed in the paper [DePlot: One-shot visual language reasoning by plot-to-table translation](https://arxiv.org/abs/2212.10505) from Fangyu Liu, Julian Martin Eisenschlos, Francesco Piccinno, Syrine Krichene, Chenxi Pang, Kenton Lee, Mandar Joshi, Wenhu Chen, Nigel Collier, Yasemin Altun.
+
+The abstract of the paper states the following:
+
+*Visual language such as charts and plots is ubiquitous in the human world. Comprehending plots and charts requires strong reasoning skills. Prior state-of-the-art (SOTA) models require at least tens of thousands of training examples and their reasoning capabilities are still much limited, especially on complex human-written queries. This paper presents the first one-shot solution to visual language reasoning. We decompose the challenge of visual language reasoning into two steps: (1) plot-to-text translation, and (2) reasoning over the translated text. The key in this method is a modality conversion module, named as DePlot, which translates the image of a plot or chart to a linearized table. The output of DePlot can then be directly used to prompt a pretrained large language model (LLM), exploiting the few-shot reasoning capabilities of LLMs. To obtain DePlot, we standardize the plot-to-table task by establishing unified task formats and metrics, and train DePlot end-to-end on this task. DePlot can then be used off-the-shelf together with LLMs in a plug-and-play fashion. Compared with a SOTA model finetuned on more than >28k data points, DePlot+LLM with just one-shot prompting achieves a 24.0% improvement over finetuned SOTA on human-written queries from the task of chart QA.*
+
+## Model description
+
+DePlot is a model that is trained using `Pix2Struct` architecture. You can find more information about `Pix2Struct` in the [Pix2Struct documentation](https://huggingface.co/docs/transformers/main/en/model_doc/pix2struct).
+DePlot is a Visual Question Answering subset of `Pix2Struct` architecture. It renders the input question on the image and predicts the answer.
+
+## Usage
+
+Currently one checkpoint is available for DePlot:
+
+- `google/deplot`: DePlot fine-tuned on ChartQA dataset 
+
+
+```python
+from transformers import AutoProcessor, Pix2StructForConditionalGeneration
+import requests
+from PIL import Image
+
+model = Pix2StructForConditionalGeneration.from_pretrained("google/deplot")
+processor = AutoProcessor.from_pretrained("google/deplot")
+url = "https://raw.githubusercontent.com/vis-nlp/ChartQA/main/ChartQA%20Dataset/val/png/5090.png"
+image = Image.open(requests.get(url, stream=True).raw)
+
+inputs = processor(images=image, text="Generate underlying data table of the figure below:", return_tensors="pt")
+predictions = model.generate(**inputs, max_new_tokens=512)
+print(processor.decode(predictions[0], skip_special_tokens=True))
+```
+
+## Fine-tuning
+
+To fine-tune DePlot, refer to the pix2struct [fine-tuning notebook](https://github.com/huggingface/notebooks/blob/main/examples/image_captioning_pix2struct.ipynb). For `Pix2Struct` models, we have found out that fine-tuning the model with Adafactor and cosine learning rate scheduler leads to faster convergence:
+```python
+from transformers.optimization import Adafactor, get_cosine_schedule_with_warmup
+
+optimizer = Adafactor(self.parameters(), scale_parameter=False, relative_step=False, lr=0.01, weight_decay=1e-05)
+scheduler = get_cosine_schedule_with_warmup(optimizer, num_warmup_steps=1000, num_training_steps=40000)
+```

docs/source/en/model_doc/matcha.mdx

@@ -0,0 +1,66 @@
+<!--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.
+-->
+
+# MatCha
+
+## Overview
+
+MatCha has been proposed in the paper [MatCha: Enhancing Visual Language Pretraining with Math Reasoning and Chart Derendering](https://arxiv.org/abs/2212.09662), from Fangyu Liu, Francesco Piccinno, Syrine Krichene, Chenxi Pang, Kenton Lee, Mandar Joshi, Yasemin Altun, Nigel Collier, Julian Martin Eisenschlos.
+
+The abstract of the paper states the following:
+
+*Visual language data such as plots, charts, and infographics are ubiquitous in the human world. However, state-of-the-art vision-language models do not perform well on these data. We propose MatCha (Math reasoning and Chart derendering pretraining) to enhance visual language models' capabilities in jointly modeling charts/plots and language data. Specifically, we propose several pretraining tasks that cover plot deconstruction and numerical reasoning which are the key capabilities in visual language modeling. We perform the MatCha pretraining starting from Pix2Struct, a recently proposed image-to-text visual language model. On standard benchmarks such as PlotQA and ChartQA, the MatCha model outperforms state-of-the-art methods by as much as nearly 20%. We also examine how well MatCha pretraining transfers to domains such as screenshots, textbook diagrams, and document figures and observe overall improvement, verifying the usefulness of MatCha pretraining on broader visual language tasks.*
+
+## Model description
+
+MatCha is a model that is trained using `Pix2Struct` architecture. You can find more information about `Pix2Struct` in the [Pix2Struct documentation](https://huggingface.co/docs/transformers/main/en/model_doc/pix2struct).
+MatCha is a Visual Question Answering subset of `Pix2Struct` architecture. It renders the input question on the image and predicts the answer.
+
+## Usage
+
+Currently 6 checkpoints are available for MatCha:
+
+- `google/matcha`: the base MatCha model, used to fine-tune MatCha on downstream tasks
+- `google/matcha-chartqa`: MatCha model fine-tuned on ChartQA dataset. It can be used to answer questions about charts.
+- `google/matcha-plotqa-v1`: MatCha model fine-tuned on PlotQA dataset. It can be used to answer questions about plots.
+- `google/matcha-plotqa-v2`: MatCha model fine-tuned on PlotQA dataset. It can be used to answer questions about plots.
+- `google/matcha-chart2text-statista`: MatCha model fine-tuned on Statista dataset. 
+- `google/matcha-chart2text-pew`: MatCha model fine-tuned on Pew dataset.
+
+The models finetuned on `chart2text-pew` and `chart2text-statista` are more suited for summarization, whereas the models finetuned on `plotqa` and `chartqa` are more suited for question answering.
+
+You can use these models as follows (example on a ChatQA dataset):
+
+```python
+from transformers import AutoProcessor, Pix2StructForConditionalGeneration
+import requests
+from PIL import Image
+
+model = Pix2StructForConditionalGeneration.from_pretrained("google/matcha-chartqa").to(0)
+processor = AutoProcessor.from_pretrained("google/matcha-chartqa")
+url = "https://raw.githubusercontent.com/vis-nlp/ChartQA/main/ChartQA%20Dataset/val/png/20294671002019.png"
+image = Image.open(requests.get(url, stream=True).raw)
+
+inputs = processor(images=image, text="Is the sum of all 4 places greater than Laos?", return_tensors="pt").to(0)
+predictions = model.generate(**inputs, max_new_tokens=512)
+print(processor.decode(predictions[0], skip_special_tokens=True))
+```
+
+## Fine-tuning
+
+To fine-tune MatCha, refer to the pix2struct [fine-tuning notebook](https://github.com/huggingface/notebooks/blob/main/examples/image_captioning_pix2struct.ipynb). For `Pix2Struct` models, we have found out that fine-tuning the model with Adafactor and cosine learning rate scheduler leads to faste convergence:
+```python
+from transformers.optimization import Adafactor, get_cosine_schedule_with_warmup
+
+optimizer = Adafactor(self.parameters(), scale_parameter=False, relative_step=False, lr=0.01, weight_decay=1e-05)
+scheduler = get_cosine_schedule_with_warmup(optimizer, num_warmup_steps=1000, num_training_steps=40000)
+```