mishig HF staff commited on
Commit
c11c5c4
1 Parent(s): de298fd

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +51 -6
README.md CHANGED
@@ -1,8 +1,53 @@
1
- ---
2
  language: en
3
  license: mit
4
- pipeline: document-question-answering
5
- tags:
6
- - layoutlm
7
- - pdf
8
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
  language: en
3
  license: mit
4
+ tags:
5
+ - layoutlm
6
+ - document-question-answering
7
+ - pdf
8
+ ---
9
+
10
+ # LayoutLM for Visual Question Answering
11
+
12
+ This is a fine-tuned version of the multi-modal [LayoutLM](https://aka.ms/layoutlm) model for the task of question answering on documents. It has been fine-tuned using both the [SQuAD2.0](https://huggingface.co/datasets/squad_v2) and [DocVQA](https://www.docvqa.org/) datasets.
13
+
14
+ ## Getting started with the model
15
+
16
+ To run these examples, you must have [PIL](https://pillow.readthedocs.io/en/stable/installation.html), [pytesseract](https://pypi.org/project/pytesseract/), and [PyTorch](https://pytorch.org/get-started/locally/) installed in addition to [transformers](https://huggingface.co/docs/transformers/index).
17
+
18
+ ```python
19
+ from transformers import pipeline
20
+
21
+ nlp = pipeline(
22
+ "document-question-answering",
23
+ model="impira/layoutlm-document-qa",
24
+ )
25
+
26
+ nlp(
27
+ "https://templates.invoicehome.com/invoice-template-us-neat-750px.png",
28
+ "What is the invoice number?"
29
+ )
30
+ # {'score': 0.9943977, 'answer': 'us-001', 'start': 15, 'end': 15}
31
+
32
+ nlp(
33
+ "https://miro.medium.com/max/787/1*iECQRIiOGTmEFLdWkVIH2g.jpeg",
34
+ "What is the purchase amount?"
35
+ )
36
+ # {'score': 0.9912159, 'answer': '$1,000,000,000', 'start': 97, 'end': 97}
37
+
38
+ nlp(
39
+ "https://www.accountingcoach.com/wp-content/uploads/2013/10/income-statement-example@2x.png",
40
+ "What are the 2020 net sales?"
41
+ )
42
+ # {'score': 0.59147286, 'answer': '$ 3,750', 'start': 19, 'end': 20}
43
+ ```
44
+
45
+ **NOTE**: This model and pipeline was recently landed in transformers via [PR #18407](https://github.com/huggingface/transformers/pull/18407) and [PR #18414](https://github.com/huggingface/transformers/pull/18414), so you'll need to use a recent version of transformers, for example:
46
+
47
+ ```bash
48
+ pip install git+https://github.com/huggingface/transformers.git@2ef774211733f0acf8d3415f9284c49ef219e991
49
+ ```
50
+
51
+ ## About us
52
+
53
+ This model was created by the team at [Impira](https://www.impira.com/).