Datasets:
de-Rodrigo
commited on
Commit
โข
40732c7
1
Parent(s):
0baebf1
Debug README
Browse files
README.md
CHANGED
@@ -121,11 +121,11 @@ tags:
|
|
121 |
<img src="figs/merit-dataset.png" alt="Visual Abstract" width="500" />
|
122 |
</p>
|
123 |
|
124 |
-
# The MERIT Dataset
|
125 |
|
126 |
The MERIT Dataset is a multimodal dataset (image + text + layout) designed for training and benchmarking Large Language Models (LLMs) on Visually Rich Document Understanding (VrDU) tasks. It is a fully labeled synthetic dataset and you can access the generation pipeline on [GitHub](https://github.com/nachoDRT/MERIT-Dataset).
|
127 |
|
128 |
-
## Introduction
|
129 |
AI faces some dynamic and technical issues that push end-users to create and gather their own data. In addition, multimodal LLMs are gaining more and more attention, but datasets to train them might be improved to be more complex, more flexible, and easier to gather/generate.
|
130 |
|
131 |
In this research project, we identify school transcripts of records as a suitable niche to generate a synthetic challenging multimodal dataset (image + text + layout) for Token Classification or Sequence Generation.
|
@@ -135,7 +135,7 @@ In this research project, we identify school transcripts of records as a suitabl
|
|
135 |
</p>
|
136 |
|
137 |
|
138 |
-
## Hardware
|
139 |
We ran the dataset generator on an MSI Meg Infinite X 10SF-666EU with an Intel Core i9-10900KF and an Nvidia RTX 2080 GPU, running on Ubuntu 20.04. Energy values in the table refer to 1k samples, and time values refer to one sample.
|
140 |
|
141 |
| Task | Energy (kWh) | Time (s) |
|
@@ -144,7 +144,7 @@ We ran the dataset generator on an MSI Meg Infinite X 10SF-666EU with an Intel C
|
|
144 |
| Modify samples in Blender | 0.366 | 34 |
|
145 |
|
146 |
|
147 |
-
## Benchmark
|
148 |
|
149 |
We train the LayoutLM family models on Token Classification to demonstrate the suitability of our dataset. The MERIT Dataset poses a challenging scenario with more than 400 labels.
|
150 |
|
|
|
121 |
<img src="figs/merit-dataset.png" alt="Visual Abstract" width="500" />
|
122 |
</p>
|
123 |
|
124 |
+
# The MERIT Dataset ๐๐๐
|
125 |
|
126 |
The MERIT Dataset is a multimodal dataset (image + text + layout) designed for training and benchmarking Large Language Models (LLMs) on Visually Rich Document Understanding (VrDU) tasks. It is a fully labeled synthetic dataset and you can access the generation pipeline on [GitHub](https://github.com/nachoDRT/MERIT-Dataset).
|
127 |
|
128 |
+
## Introduction โน๏ธ
|
129 |
AI faces some dynamic and technical issues that push end-users to create and gather their own data. In addition, multimodal LLMs are gaining more and more attention, but datasets to train them might be improved to be more complex, more flexible, and easier to gather/generate.
|
130 |
|
131 |
In this research project, we identify school transcripts of records as a suitable niche to generate a synthetic challenging multimodal dataset (image + text + layout) for Token Classification or Sequence Generation.
|
|
|
135 |
</p>
|
136 |
|
137 |
|
138 |
+
## Hardware โ๏ธ
|
139 |
We ran the dataset generator on an MSI Meg Infinite X 10SF-666EU with an Intel Core i9-10900KF and an Nvidia RTX 2080 GPU, running on Ubuntu 20.04. Energy values in the table refer to 1k samples, and time values refer to one sample.
|
140 |
|
141 |
| Task | Energy (kWh) | Time (s) |
|
|
|
144 |
| Modify samples in Blender | 0.366 | 34 |
|
145 |
|
146 |
|
147 |
+
## Benchmark ๐ช
|
148 |
|
149 |
We train the LayoutLM family models on Token Classification to demonstrate the suitability of our dataset. The MERIT Dataset poses a challenging scenario with more than 400 labels.
|
150 |
|