Add GOT-OCR 2.0 to Transformers #34721
yonigozlan
merged
53 commits into
huggingface:main
from
yonigozlan:add-got-ocr2
Ucas-HaoranWei
approved these changes
on 2024-11-15
Ucas-HaoranWei
approved these changes
on 2024-11-15
Ucas-HaoranWei
approved these changes
on 2024-11-15
Ucas-HaoranWei
approved these changes
on 2024-11-18
init modular got_ocr2
cede6406
Get correct got_ocr architecture
72a003c3
add processing
7bda0529
run modular with processing
fdba2c56
add working inference
5251b05d
apply modular
fc64947c
Refactor and fix style
c8352b41
Refactor, cleanup, fix style
69fca747
fix init order
21effc48
Fix docs
fb9f006d
add base modeling tests
ca53c398
fix style and consistency
73da47e1
rename doc file
c7696353
fix repo consistency
fc49a527
fix inference with box
d546ac87
add image processing and support for crop_to_multi_page
66951798
Fix batch inference
fdbf0d20
add tests
764d456a
fixup
782e38cd
fix slow test
fea6929c
fix docstrings
b094771c
Add model doc
9ab7d9e2
update to new init
a843c4e6
fix input autocast pixel_values dtype
ee2b99ab
update doc
8fe8edeb
move doc to multimodal
4785527e
Reformat crop_image_to_patches and add docstrings
5d81df1f
Fix example in forward docstring
8784bcf5
Address Pablo review
a312ca2e
[run slow] got_ocr2
5500203c
remove defaults defined twice
f7596084
apply modular
7cea6a24
add torch_device to integration tests
3ef43eef
update modular
088672f1
follow-up Pavel review
7e6bab96
add device variable in doc
1f5f054c
fix doc multi-page
0ff44e8e
Force eager attention for vision encoder to avoid attn implementation…
3ae43ec5
ArthurZucker
commented
on 2025-01-23
Merge remote-tracking branch 'upstream/main' into add-got-ocr2
8289e69c
revert qwen2vl doc changes
9cbd8049
use Qwen2ForCausalLM instead of Qwen2Model
c87fa624
make fixup
34c716ed
refactor gotocr2 to llava style
0178c681
uniformize function names and reduce checks
31204126
Merge branch 'main' into add-got-ocr2
6b5169b4
Merge branch 'main' into add-got-ocr2
3b626beb
final nits
b5b56d86
Merge branch 'add-got-ocr2' of https://github.com/yonigozlan/transfor…
991b3f37
Merge branch 'main' into add-got-ocr2
9046bf54
fix pixel_values dtype error
cc861d3e
change checkpoint names
506d5665
Merge branch 'main' into add-got-ocr2
2dbbf24e
fix modular
9fb4e909
Reviewers
Assignees
No one assigned
Labels
New model
Multimodal
run-slow
Milestone
No milestone
Login to write a write a comment.
Login via GitHub