9iCSSKJrJr SSKJrJrJr SSKJr SSK J r SSK J r \ "5r \Sr\\\\4\4r"SS\5rg ) )ABCabstractmethod)AnyDictUnion)Model) ConfigDict) get_logger)z torch.Tensorz tf.Tensorcb\rSrSrSrSr\S\\\ 44Sj5r \S\\\ 44Sj5r Sr g) HeadzAThe head base class is for the tasks head method definition c $[U5Ulg)N)r config)selfkwargss `/var/www/html/land-doc-ocr/venv/lib/python3.13/site-packages/modelscope/models/base/base_head.py__init__ Head.__init__s ( returncg)z This method will use the output from backbone model to do any downstream tasks. Receive The output from backbone model. Returns (Dict[str, Any]): The output from downstream task. Nrargsrs rforward Head.forwards rcg)zW compute loss for head during the finetuning. Returns (Dict[str, Any]): The loss dict Nrrs r compute_lossHead.compute_loss!s r)rN) __name__ __module__ __qualname____firstlineno____doc__rrrstrrrr__static_attributes__rrrr r sR) $sCx.   tCH~  rr N)abcrrtypingrrr!modelscope.models.base.base_modelrmodelscope.utils.configr modelscope.utils.loggerr loggerTensorr%Inputr rrrr/sM###3..  * + d3;&' 3 r