o ίi7 @sXdZddlmZmZddlmZddlmZmZm Z m Z m Z ddl m Z d@dedeefd d ZeGd d d ZeGd ddZiZiZedddddddddd ed<edddddded<edddddd ddd!d ed"<ed#ddddded"<ed$dd%d&d&d ddd!d ed'<ed"ed'<ed(dd(d&d)d dd*d!d ed+<ed#d*d,d&d*ded+<ed(dd-d&d.d dd*d!d ed/<ed+ed/<eeddd0ddd1d2d3ed4<eed"dd0ddd1d5d3ed6<ed4ed7<ed6ed8<eeddd0ddd1d9ed:<eed"dd0ddd;ed<<eed'd*d0ddd;ed=<eed+d*d0ddd;ed><eed/d*d0ddd;ed?<dS)Az6 Include all available vision encoder configurations. ) dataclassreplace)partial)CallableOptionalSequenceTupleList)hf_hub_downloadNnamepathcCsN|p d|d|d}|dr%|tdd}|d\}}t||dS|S)Nzhf://facebook/:z.ptzhf://)repo_idfilename) startswithlensplitr )r r repofilerN/home/ubuntu/.local/lib/python3.10/site-packages/core/vision_encoder/config.pyfetch_pe_checkpoints   rc@seZdZUdZeed<eed<eed<eed<eed<eeed<dZeed <d Z eed <d Z eed <dZ e ed<dZ e ed<dZe ed<dZeed<dZeed<dZe ed<dZe ed<dS)PEConfigz Vision Tower Config. patch_sizewidthlayersheads mlp_ratio output_dimN ls_init_valueg drop_path) image_sizeTuse_abs_posembF use_cls_token use_rope2dattn pool_typeattn_pooler_heads use_ln_pre use_ln_post)__name__ __module__ __qualname____doc__int__annotations__floatrrr r"r#boolr$r%r'strr)r*r+rrrrrs$           rc@sReZdZUdZeed<eed<eed<eed<eed<dZeed<d Zeed <d S) PETextConfigz Text Tower Config. context_lengthrrrr@ri vocab_sizeN) r,r-r.r/r0r1rr2r8rrrrr58s  r5ii2gUUUUUU@r&iF) r"rrrrrr'rr$zPE-Core-G14-448H)r6rrrriPir7TzPE-Core-L14-336 r!i zPE-Core-B16-224iizPE-Core-S16-384r(zPE-Core-T16-384noneg?/)r"r'r+rrrzPE-Lang-G14-448zPE-Lang-L14-448zPE-Lang-G14-448-TilingzPE-Lang-L14-448-Tiling)r"r'r+rrzPE-Spatial-G14-448)r"r'r+rzPE-Spatial-L14-448zPE-Spatial-B16-512zPE-Spatial-S16-512zPE-Spatial-T16-512)N)r/ dataclassesrr functoolsrtypingrrrrr huggingface_hubr r4rrr5PE_VISION_CONFIGPE_TEXT_CONFIGrrrrs$