Views
No views yet
1from transformers import XLNetTokenizer, TFGPT2LMHeadModel
2import jieba
3
4# add spicel process
5class XLNetTokenizer(XLNetTokenizer):
6 translator = str.maketrans(" \n", "\u2582\u2583")
7
8 def _tokenize(self, text, *args, **kwargs):
9 text = [x.translate(self.translator) for x in jieba.cut(text, cut_all=False)]
10 text = " ".join(text)
11 return super()._tokenize(text, *args, **kwargs)
12
13 def _decode(self, *args, **kwargs):
14 text = super()._decode(*args, **kwargs)
15 text = text.replace(' ', '').replace('\u2582', ' ').replace('\u2583', '\n')
16 return text
17
18
19tokenizer = XLNetTokenizer.from_pretrained('mymusise/CPM-GPT2-FP16')
20model = TFGPT2LMHeadModel.from_pretrained("mymusise/CPM-GPT2-FP16")1from transformers import TextGenerationPipeline
2
3
4text_generater = TextGenerationPipeline(model, tokenizer)
5
6texts = [
7 '今天天气不错',
8 '天下武功, 唯快不',
9 """
10 我们在火星上发现了大量的神奇物种。有神奇的海星兽，身上是粉色的，有5条腿；有胆小的猫猫兽，橘色，有4条腿；有令人恐惧的蜈蚣兽，全身漆黑，36条腿；有纯洁的天使兽，全身洁白无瑕，有3条腿；有贪吃的汪汪兽，银色的毛发，有5条腿；有蛋蛋兽，紫色，8条腿。
11
12 请根据上文，列出一个表格，包含物种名、颜色、腿数量。
13 |物种名|颜色|腿数量|
14 |亚古兽|金黄|2|
15 |海星兽|粉色|5|
16 |猫猫兽|橘色|4|
17 |蜈蚣兽|漆黑|36|
18 """
19]
20
21for text in texts:
22 token_len = len(tokenizer._tokenize(text))
23 print(text_generater(text, max_length=token_len + 15, top_k=1, use_cache=True, prefix='')[0]['generated_text'])
24 print(text_generater(text, max_length=token_len + 15, do_sample=True, top_k=5)[0]['generated_text'])
