add Qianfan-OCR model definition (#45280)
* add QianfanOCR Model
* update remote model id
* update model attributes
* fix errors in ddp
* fix dtype mismatch
* fix template error
* fix style
* fix code style
* Update expected logits and output strings in tests (tested on 4090)
* update expected result under 4090
* align test result on 4090 x2
* fix CI run_tests issue
* add QianfanOCR Model
* fix CI run_tests issue
* using conversion mapping to mapping weights
* fix ddp issue
* fix processor UT errors
* squeeze modular to make use of existing implementation
* update format fixs by tools
* Apply suggestion from @vasqu
Co-authored-by: Anton Vlasjuk <73884904+vasqu@users.noreply.github.com>
* Update src/transformers/models/qianfan_ocr/modular_qianfan_ocr.py
Co-authored-by: Anton Vlasjuk <73884904+vasqu@users.noreply.github.com>
* fix format/code style issue from comments
* fix UT errors
* fix processor UT errors
* formatting code
* refactor modular
* refactor test to use VLMModelTest
* format code
* fix data in doc
* remove compile graph
* fix modular inheritance on config
* no need to skip
* fix auto mappings and apply fix repo
* fix
* fix
* nits
* move conversion mapping
* fix integration test for our CI
---------
Co-authored-by: zhuangbairong <zhuangbairong@baidu.com>
Co-authored-by: Anton Vlasjuk <73884904+vasqu@users.noreply.github.com>
Co-authored-by: vasqu <antonprogamer@gmail.com>
