add Qianfan-OCR model definition #45280
vasqu
merged
40 commits into
huggingface:main
from
marvinzh:add-qianfan-ocr
add QianfanOCR Model
34694a2e
update remote model id
08fe5fae
update model attributes
14c9187e
fix errors in ddp
d4ec4c0a
fix dtype mismatch
4e3060a9
fix template error
c1871817
fix style
c098b821
fix code style
82fd32ea
Update expected logits and output strings in tests (tested on 4090)
5578485d
update expected result under 4090
78386a0b
align test result on 4090 x2
6c4902e5
fix CI run_tests issue
572634c9
add QianfanOCR Model
f006bad2
fix CI run_tests issue
7df2d896
using conversion mapping to mapping weights
98b48d9e
fix ddp issue
105da02f
fix processor UT errors
6181828d
squeeze modular to make use of existing implementation
391f9d22
update format fixs by tools
9b834d8a
Apply suggestion from @vasqu
5bd2d12f
Update src/transformers/models/qianfan_ocr/modular_qianfan_ocr.py
a98a1fca
fix format/code style issue from comments
ca747c22
fix UT errors
cedf73ba
fix processor UT errors
683f6860
formatting code
58ae5fa1
refactor modular
d67c5ff0
refactor test to use VLMModelTest
eeb4efbd
format code
58755e40
fix data in doc
c9e88e81
remove compile graph
1d218c75
fix modular inheritance on config
92fa1c3c
no need to skip
b59e6b84
Merge branch 'main' into add-qianfan-ocr
ade2eff5
fix auto mappings and apply fix repo
a4a6f503
fix
36026a52
fix
4edb19e0
nits
45897a84
move conversion mapping
cfd2a9cc
Merge branch 'main' into add-qianfan-ocr
07e75381
fix integration test for our CI
2fbd0d7a
Login to write a write a comment.
Login via GitHub