a g @s>dZddlmZddlmZmZddlmZGdddZdS)z Summarization by mT5 model )List) T5TokenizerMT5ForConditionalGeneration)CPE_KMUTT_THAI_SENTENCE_SUMc @s<eZdZdeeeeeeedd d Zeeed d d ZdS) mT5SummarizersmalldTN) model_size num_beamsno_repeat_ngram_size min_length max_lengthskip_special_tokenspretrained_mt5_model_namec Csd}|dur0|dvr$td|dd|}n|tkrDdt}n|}||_t||_t||_||_||_ ||_ ||_ ||_ dS)N)rbaseZlargeZxlZxxlz model_size "zY" not found. It might be a typo; if not, please consult our document.z google/mt5- thanathorn/) ValueErrorr model_namerZfrom_pretrainedmodelr tokenizerr rrrr) selfr r rrrrrrrA/usr/local/lib/python3.9/dist-packages/pythainlp/summarize/mt5.py__init__s$      zmT5Summarizer.__init__)textreturncCs~|dd}|jdtkr*d|}nd|}|jj|dd}|jj||j|j |j |j dd }|jj |d |j d }|gS) N rrz simplify: z summarize: pt)Zreturn_tensorsT)r rrrZearly_stoppingr)r)stripreplacerrrencodergenerater rrrdecoder)rrZpreprocess_textZt5_prepared_TextZtokenized_textZ summary_idsoutputrrr summarize.s&  zmT5Summarizer.summarize)rrr r r TN) __name__ __module__ __qualname__strintboolrrr(rrrrr s" rN) __doc__typingrZ transformersrrZpythainlp.summarizerrrrrrs