RUPunct/RUPunct_big

1---2license: mit3language:4- ru5---6 7RUPunct_big - самая большая модель из семейства RUPunct. Подходит для большинства задач. 8 9Код инференса:10```py11from transformers import pipeline12from transformers import AutoTokenizer13 14pt = "RUPunct/RUPunct_big"15 16tk = AutoTokenizer.from_pretrained(pt, strip_accents=False, add_prefix_space=True)17classifier = pipeline("ner", model=pt, tokenizer=tk, aggregation_strategy="first")18 19 20def process_token(token, label):21    if label == "LOWER_O":22        return token23    if label == "LOWER_PERIOD":24        return token + "."25    if label == "LOWER_COMMA":26        return token + ","27    if label == "LOWER_QUESTION":28        return token + "?"29    if label == "LOWER_TIRE":30        return token + "—"31    if label == "LOWER_DVOETOCHIE":32        return token + ":"33    if label == "LOWER_VOSKL":34        return token + "!"35    if label == "LOWER_PERIODCOMMA":36        return token + ";"37    if label == "LOWER_DEFIS":38        return token + "-"39    if label == "LOWER_MNOGOTOCHIE":40        return token + "..."41    if label == "LOWER_QUESTIONVOSKL":42        return token + "?!"43    if label == "UPPER_O":44        return token.capitalize()45    if label == "UPPER_PERIOD":46        return token.capitalize() + "."47    if label == "UPPER_COMMA":48        return token.capitalize() + ","49    if label == "UPPER_QUESTION":50        return token.capitalize() + "?"51    if label == "UPPER_TIRE":52        return token.capitalize() + " —"53    if label == "UPPER_DVOETOCHIE":54        return token.capitalize() + ":"55    if label == "UPPER_VOSKL":56        return token.capitalize() + "!"57    if label == "UPPER_PERIODCOMMA":58        return token.capitalize() + ";"59    if label == "UPPER_DEFIS":60        return token.capitalize() + "-"61    if label == "UPPER_MNOGOTOCHIE":62        return token.capitalize() + "..."63    if label == "UPPER_QUESTIONVOSKL":64        return token.capitalize() + "?!"65    if label == "UPPER_TOTAL_O":66        return token.upper()67    if label == "UPPER_TOTAL_PERIOD":68        return token.upper() + "."69    if label == "UPPER_TOTAL_COMMA":70        return token.upper() + ","71    if label == "UPPER_TOTAL_QUESTION":72        return token.upper() + "?"73    if label == "UPPER_TOTAL_TIRE":74        return token.upper() + " —"75    if label == "UPPER_TOTAL_DVOETOCHIE":76        return token.upper() + ":"77    if label == "UPPER_TOTAL_VOSKL":78        return token.upper() + "!"79    if label == "UPPER_TOTAL_PERIODCOMMA":80        return token.upper() + ";"81    if label == "UPPER_TOTAL_DEFIS":82        return token.upper() + "-"83    if label == "UPPER_TOTAL_MNOGOTOCHIE":84        return token.upper() + "..."85    if label == "UPPER_TOTAL_QUESTIONVOSKL":86        return token.upper() + "?!"87 88while 1:89    input_text = input(":> ")90    preds = classifier(input_text)91    output = ""92    for item in preds:93        output += " " + process_token(item['word'].strip(), item['entity_group'])94    print(">>>", output)95```