[Model] Support Unlimited OCR (#46564)

Signed-off-by: Tianyu Guo <guoty9@mail2.sysu.edu.cn>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Roger Wang <hey@rogerw.io>
This commit is contained in:
Tianyu Guo
2026-06-27 23:09:18 -07:00
committed by GitHub
co-authored by Isotr0py Roger Wang
parent a65f93fb2e
commit c6741b2ad4
35 changed files with 1084 additions and 49 deletions
+5 -1
View File
@@ -31,7 +31,11 @@ logger = init_logger(__name__)
# temporary workaround and better long term solutions are:
# - Add model type to MODELS_WITH_INCORRECT_HUB_TOKENIZER_CLASS in transformers (better)
# - Fix tokenizer_class on the hub for the affected models (best)
_MODEL_TYPES_WITH_INCORRECT_TOKENIZER_CLASS: set[str] = {"step3_vl", "step3p7"}
_MODEL_TYPES_WITH_INCORRECT_TOKENIZER_CLASS: set[str] = {
"step3_vl",
"step3p7",
"unlimited-ocr",
}
_VLLM_TOKENIZERS = {
"deepseek_v32": ("deepseek_v32", "DeepseekV32Tokenizer"),