比较与torchtext.data.functional.sentencepiece_tokenizer的差异

torchtext.data.functional.sentencepiece_tokenizer

torchtext.data.functional.sentencepiece_tokenizer(
    sp_model
)

mindspore.dataset.text.SentencePieceTokenizer

class mindspore.dataset.text.SentencePieceTokenizer(
    mode,
    out_type
)

更多内容详见mindspore.dataset.text.SentencePieceTokenizer。

使用方式

PyTorch：依据传入的分词模型，返回将文本转换为字符串的生成器。

MindSpore：依据传入的分词模型，对输入的文本进行分词及标记；输出类型是string或int类型。

分类	子类	PyTorch	MindSpore	差异
参数	参数1	sp_model	mode	MindSpore支持SentencePiece词汇表或分词模型地址
	参数2	-	out_type	分词器输出的类型

代码示例

from download import download

url = "https://mindspore-website.obs.cn-north-4.myhuaweicloud.com/notebook/datasets/sentencepiece.bpe.model"
download(url, './sentencepiece.bpe.model', replace=True)

# PyTorch
from torchtext.data.functional import load_sp_model, sentencepiece_tokenizer

list_a = "sentencepiece encode as pieces"
model = load_sp_model("./sentencepiece.bpe.model")
sp_id_generator = sentencepiece_tokenizer(model)
print(list(sp_id_generator([list_a])))
# Out: [['▁sentence', 'piece', '▁en', 'code', '▁as', '▁pieces']]

# MindSpore
import mindspore.dataset.text as text

sp_id_generator = text.SentencePieceTokenizer("./sentencepiece.bpe.model", out_type=text.SPieceTokenizerOutType.STRING)
print(list(sp_id_generator(list_a)))
# Out: ['▁sentence', 'piece', '▁en', 'code', '▁as', '▁pieces']