optimum-intel-internal-testing/tiny-random-minicpmv-2_6
129k
1from transformers.models.qwen2 import Qwen2TokenizerFast2 3 4class MiniCPMVTokenizerFast(Qwen2TokenizerFast):5 def __init__(self, **kwargs):6 super().__init__(**kwargs)7 self.im_start = "<image>"8 self.im_end = "</image>"9 self.ref_start = "<ref>"10 self.ref_end = "</ref>"11 self.box_start = "<box>"12 self.box_end = "</box>"13 self.quad_start = "<quad>"14 self.quad_end = "</quad>"15 self.slice_start = "<slice>"16 self.slice_end = "</slice>"17 self.im_id_start = "<image_id>"18 self.im_id_end = "</image_id>"19 20 @property21 def eos_id(self):22 return self.eos_token_id23 24 @property25 def bos_id(self):26 return self.bos_token_id27 28 @property29 def unk_id(self):30 return self.unk_token_id31 32 @property33 def im_start_id(self):34 return self.convert_tokens_to_ids(self.im_start)35 36 @property37 def im_end_id(self):38 return self.convert_tokens_to_ids(self.im_end)39 40 @property41 def slice_start_id(self):42 return self.convert_tokens_to_ids(self.slice_start)43 44 @property45 def slice_end_id(self):46 return self.convert_tokens_to_ids(self.slice_end)47 48 @property49 def im_id_start_id(self):50 return self.convert_tokens_to_ids(self.im_id_start)51 52 @property53 def im_id_end_id(self):54 return self.convert_tokens_to_ids(self.im_id_end)55 56 @property57 def newline_id(self):58 return self.convert_tokens_to_ids("\n")59 60 @staticmethod61 def escape(text: str) -> str:62 return text63 64 @staticmethod65 def unescape(text: str) -> str:66 return text67 