
!pip install transformers[ja]

---------------------------------------------------------------

Xg1

#  p
from transformers import pipeline

nlp = pipeline("sentiment-analysis")

print(nlp("I love you"))
print(nlp("I hate you"))

---------------------------------------------------------------

Xg2

#  {
nlp = pipeline("sentiment-analysis",
               model="daigo/bert-base-japanese-sentiment",
               tokenizer="daigo/bert-base-japanese-sentiment")

print(nlp(" ̏i𔃂Ă悩B"))
print(nlp(" ̏i͔ĎsB"))

---------------------------------------------------------------

ύX

print(nlp(" ̃[͍DB"))
print(nlp(" ̃[͍Dł͂ȂB"))

---------------------------------------------------------------

Xg3

#  p
nlp = pipeline("fill-mask")

results = nlp(f"I eat {nlp.tokenizer.mask_token} everyday.")
for result in results:
  print(result)

---------------------------------------------------------------

Xg4

#  {
from transformers import BertConfig, AutoTokenizer, BertForMaskedLM

config = BertConfig.from_pretrained(
    'cl-tohoku/bert-base-japanese-whole-word-masking')
tokenizer = AutoTokenizer.from_pretrained(
    'cl-tohoku/bert-base-japanese-whole-word-masking')
model = BertForMaskedLM.from_pretrained(
    'cl-tohoku/bert-base-japanese-whole-word-masking')

nlp = pipeline('fill-mask', model=model,
               tokenizer=tokenizer, config=config)
results = nlp(' H[MASK] HׂB')

for result in results:
  print(result)

---------------------------------------------------------------

ύX

[т[MASK] HׂB


---------------------------------------------------------------

ύX

LaHׂȂ̂́Aꂪ[MASK] B

---------------------------------------------------------------

Xg5

text1 = ' LaHׂȂB'
ids = tokenizer.encode(text1, return_tensors='pt')

print(tokenizer.convert_ids_to_tokens(ids[0].tolist()))


ȏ