NchuNLP/Chinese-Question-Answering

from transformers import BertTokenizerFast, BertForQuestionAnswering, pipeline
model_name = "NchuNLP/Chinese-Question-Answering"
tokenizer = BertTokenizerFast.from_pretrained(model_name)
model = BertForQuestionAnswering.from_pretrained(model_name)

# a) Obtener predicciones
nlp = pipeline('question-answering', model=model, tokenizer=tokenizer)
QA_input = {
'question': '中興大學在哪裡?',
'context': '國立中興大學(簡稱興大、NCHU),是位於臺中的一所高等教育機構。中興大學以農业科學、農業經濟學、獸ī學、生命科學、轉譯醫學、生ī工程、生物科技、綠ī科技等研究領域見長 。近年中興大學與臺中榮民總醫院、彰化師範大學、中國醫藥大學等機構合作,聚焦於癌症醫學、免疫醫學及醫學ī工程三項領域,將實驗室成果逐步應用到臨床上,未來「衛生福利部南投醫院中興院區」將改為「國立中興大學醫學院附設醫院」。興大也與臺中市政府合作,簽訂合作意向書,共同推動數位文化、智慧城市等面相ī動區域發展。'}
res = nlp(QA_input)

{'score': 1.0, 'start': 21, 'end': 23, 'answer': '臺中'}

# b) Dentro del pipeline de Respuesta a Preguntas

inputs = tokenizer(query, text, return_tensors="pt",padding=True, truncation=True, max_length=512, stride=256)
outputs = model(**inputs)

sequence_ids = inputs.sequence_ids()
# Enmascarar todo menos los tokens del contexto
mask = [i != 1 for i in sequence_ids]
# Desenmascarar el token [CLS]
mask[0] = False
mask = torch.tensor(mask)[None]

start_logits[mask] = -10000
end_logits[mask] = -10000

start_probabilities = torch.nn.functional.softmax(start_logits, dim=-1)[0]
end_probabilities = torch.nn.functional.softmax(end_logits, dim=-1)[0]

scores = start_probabilities[:, None] * end_probabilities[None, :]

max_index = scores.argmax().item()
start_index = max_index // scores.shape[1]
end_index = max_index % scores.shape[1]


inputs_with_offsets = tokenizer(query, text, return_offsets_mapping=True)
offsets = inputs_with_offsets["offset_mapping"]

start_char, _ = offsets[start_index]
_, end_char = offsets[end_index]
answer = text[start_char:end_char]

result = {
"answer": answer,
"start": start_char,
"end": end_char,
"score": scores[start_index, end_index],
}
print(result)