Add Nougat #25942
ArthurZucker
merged
83 commits into
huggingface:main
from
NielsRogge:add_nougat
NielsRogge
commented
on 2023-09-17
Add conversion script
8a48f52a
Add NougatImageProcessor
80feb49e
Add crop margin
b4612d80
More improvements
f9f88116
Add docs, READMEs
5241d13b
Remove print statements
75409b7b
Include model_max_length
4ae415aa
Add NougatTokenizerFast
0d28eeab
Fix imports
9397bf43
Improve postprocessing
8fd62989
Improve image processor
1d5a48d0
Fix image processor
98ac3e8d
Improve normalize method
85152620
More improvements
111db34e
More improvements
f846e4e8
Add processor, improve docs
6de38a49
Simplify fast tokenizer
5580b623
Remove test file
b8701cee
Fix docstrings
832c7734
Use NougatProcessor in conversion script
e2282833
Add is_levensthein_available
e6e07fe5
Add tokenizer tests
dca7ce85
More improvements
5d6e186a
Use numpy instead of opencv
40847a55
Add is_cv2_available
f425717a
Fix cv2_available
a736aa80
Add is_nltk_available
6d03accc
Add image processor tests, improve crop_margin
d2384ee2
Add integration tests
eff1f832
Improve integration test
50bbed82
Use do_rescale instead of hacks, thanks Amy
7149570f
Remove random_padding
a430504d
Address comments
75b30658
Address more comments
61afed32
Add import
4e2d73c5
Address more comments
6bedc697
Address more comments
cc7b8a4c
Address comment
45c46692
Address comment
dca63691
Set max_model_input_sizes
528bf157
Add tests
46ee09c9
Add requires_backends
771be65e
Add Nougat to exotic tests
a41ad761
Use to_pil_image
5575bdd1
Address comment regarding nltk
b9c83ba8
Add NLTK
5b967f9f
Improve variable names, integration test
0b4ad48e
Add test
e8e36d88
refactor, document, and test regexes
5c53a694
remove named capture groups, add comments
1b9599df
format
f4354434
add non-markdown fixed tokenization
8832d80b
format
1809ccf5
correct flakyness of args parse
d5468f8d
add regex comments
cb68a4e9
test functionalities for crop_image, align long axis and expected output
87c7c88c
add regex tests
11edb666
remove cv2 dependency
d27c64aa
test crop_margin equality between cv2 and python
a8f11548
refactor table regexes to markdown
75d9801e
change print to log, improve doc
6ad7f381
fix high count tables correction
ef89a5e9
address PR comments: naming, linting, asserts
02cc7858
Address comments
be9fe548
Add copied from
8b9e291f
Update conversion script
cf258a8f
Update conversion script to convert both small and base versions
c69b0a03
Add inference example
06dffc4c
Add more info
e558513e
Fix style
bd715dcb
Add require annotators to test
b99439d5
Define all keyword arguments explicitly
6b1fdefc
Move cv2 annotator
5fbfe1b9
Add tokenizer init method
7c94c862
Transfer checkpoints
34beb164
Add reference to Donut
93d7055a
Address comments
9f304aa2
Skip test
fea2a16d
Remove cv2 method
86c4567e
Add copied from statements
062a14d5
Use cached_property
313097a7
Fix docstring
92759af4
Add file to not doctested
d5e05900
LysandreJik
approved these changes
on 2023-09-25
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub