OpenFormosa/PangolinTokenizer
312
1{2 "unk_token": "<unk>",3 "bos_token": "<s>",4 "eos_token": "</s>",5 "pad_token": "<pad>",6 "additional_special_tokens": [7 "<|bos|>",8 "<|eos|>",9 "<|unk|>",10 "<|pad|>",11 "<|system|>",12 "<|user_channel|>",13 "<|assistant_channel|>",14 "<|task:speech_to_text|>",15 "<|task:text_to_speech|>",16 "<|input_audio_start|>",17 "<|input_audio_end|>",18 "<|audio_ref_start|>",19 "<|audio_ref_end|>",20 "<|audio_start|>",21 "<|audio_end|>",22 "<|speech_start|>",23 "<|speech_end|>",24 "<|transcript_start|>",25 "<|transcript_end|>",26 "<|segment_start|>",27 "<|segment_end|>",28 "<|speaker|>",29 "<|start_time|>",30 "<|end_time|>",31 "<|duration|>",32 "<|content|>",33 "<|non_speech_event|>",34 "<|retrieval_result_start|>",35 "<|retrieval_result_end|>",36 "<|ocr_start|>",37 "<|ocr_end|>",38 "<|image_start|>",39 "<|image_end|>",40 "<|video_start|>",41 "<|video_end|>",42 "<|user|>",43 "<|assistant|>",44 "<|tool_call|>",45 "<|tool_response|>",46 "<|endoftext|>",47 "<think>",48 "</think>",49 "<|no_think|>",50 "<|think|>",51 "<|think_max|>",52 "<|task:text_to_text|>",53 "<|task:speech_to_speech|>",54 "<|task:text_speech_to_text|>",55 "<|task:text_speech_to_speech|>",56 "<|task:full_duplex_speech|>",57 "<|task:agent|>",58 "<|task:tool_use|>",59 "<|task:rag|>",60 "<|task:code_execution|>",61 "<|task:document_qa|>",62 "<|task:data_analysis|>",63 "<|task:workflow|>",64 "<|reasoning_mode:none|>",65 "<|reasoning_mode:short|>",66 "<|reasoning_mode:deep|>",67 "<|reasoning_mode:verify|>",68 "<|private_reasoning_start|>",69 "<|private_reasoning_end|>",70 "<|reasoning_summary_start|>",71 "<|reasoning_summary_end|>",72 "<|plan_start|>",73 "<|plan_end|>",74 "<|step_start|>",75 "<|step_end|>",76 "<|action_start|>",77 "<|action_end|>",78 "<|observation_start|>",79 "<|observation_end|>",80 "<|reflection_start|>",81 "<|reflection_end|>",82 "<|verification_start|>",83 "<|verification_end|>",84 "<|tool_schema_start|>",85 "<|tool_schema_end|>",86 "<|tool_call_start|>",87 "<|tool_call_end|>",88 "<|tool_result_start|>",89 "<|tool_result_end|>",90 "<|tool_error_start|>",91 "<|tool_error_end|>",92 "<|retrieval_query_start|>",93 "<|retrieval_query_end|>",94 "<|citation_start|>",95 "<|citation_end|>",96 "<|memory_read_start|>",97 "<|memory_read_end|>",98 "<|memory_write_start|>",99 "<|memory_write_end|>",100 "<|final_answer_start|>",101 "<|final_answer_end|>",102 "<|json_start|>",103 "<|json_end|>",104 "<|code_start|>",105 "<|code_end|>",106 "<|markdown_start|>",107 "<|markdown_end|>",108 "<|duplex_start|>",109 "<|duplex_end|>",110 "<|system_channel|>",111 "<|listen|>",112 "<|speak|>",113 "<|listen_speak|>",114 "<|output_audio_start|>",115 "<|output_audio_end|>",116 "<|text_start|>",117 "<|text_end|>",118 "<|overlap|>",119 "<|barge_in|>",120 "<|interruption|>",121 "<|interruption_repair|>",122 "<|backchannel|>",123 "<|turn_yield|>",124 "<|hold|>",125 "<|silence|>",126 "<|non_speech|>",127 "<|voice_reference_start|>",128 "<|voice_reference_end|>",129 "<|voice_reference|>",130 "<|voice_switch|>",131 "<|speaker_style|>",132 "<|prosody_control|>",133 "<|zh_tw|>",134 "<|zh_hant|>",135 "<|taigi|>",136 "<|hakka|>",137 "<|bopomofo|>",138 "<|mixed_en|>",139 "<|en|>",140 "<|ja|>",141 "<|ko|>",142 "<|vi|>",143 "<|id|>",144 "<|th|>",145 "<|asr|>",146 "<|tts|>",147 "<|speaker_0|>",148 "<|speaker_1|>",149 "<|speaker_2|>",150 "<|speaker_3|>",151 "<|timestamp|>",152 "<|noise|>",153 "<|laugh|>",154 "<|breath|>",155 "<|pause|>",156 "<|prosody|>",157 "<|pron|>",158 "</|pron|>",159 "<|image|>",160 "<|ocr|>",161 "<|bbox|>",162 "<|line|>",163 "<|table|>",164 "<|row|>",165 "<|col|>",166 "<|cell|>",167 "<|reading_order|>",168 "<|source|>",169 "<|cite|>",170 "<|evidence|>",171 "<|quote|>"172 ]173}174 