语言:

• 英语 标签:
• 语法错误修正
• 文本到文本
• T5 许可证: Apache-2.0 数据集:
• clang-8
• conll-14
• conll-13 评估指标:
• F0.5

该模型是谷歌论文《多语言语法错误修正的简易方案》(https://arxiv.org/pdf/2106.03830.pdf)的实现，该论文在语法错误修正(GEC)任务中报告了最先进的分数。我们实现了T5-small版本，其F_0.5分数与论文中报告的一致(60.70)。

要有效使用"托管推理API"，请写入"gec: [您的句子]"。

使用该模型的示例如下：

from transformers import T5ForConditionalGeneration, T5Tokenizer

model = T5ForConditionalGeneration.from_pretrained("Unbabel/gec-t5_small")
tokenizer = T5Tokenizer.from_pretrained('t5-small')

sentence = "I like to swimming"
tokenized_sentence = tokenizer('gec: ' + sentence, max_length=128, truncation=True, padding='max_length', return_tensors='pt')
corrected_sentence = tokenizer.decode(
    model.generate(
        input_ids = tokenized_sentence.input_ids,
        attention_mask = tokenized_sentence.attention_mask, 
        max_length=128,
        num_beams=5,
        early_stopping=True,
    )[0],
    skip_special_tokens=True, 
    clean_up_tokenization_spaces=True
)
print(corrected_sentence) # -> 输出：I like swimming.