Views
No views yet
1import torch
2from transformers import T5Tokenizer, GPT2LMHeadModel
3
4tokenizer = T5Tokenizer.from_pretrained("skytnt/gpt2-japanese-lyric-small")
5model = GPT2LMHeadModel.from_pretrained("skytnt/gpt2-japanese-lyric-small")
6
7
8def gen_lyric(prompt_text: str):
9 prompt_text = "<s>" + prompt_text.replace("\n", "\\n ")
10 prompt_tokens = tokenizer.tokenize(prompt_text)
11 prompt_token_ids = tokenizer.convert_tokens_to_ids(prompt_tokens)
12 prompt_tensor = torch.LongTensor(prompt_token_ids).to(device)
13 prompt_tensor = prompt_tensor.view(1, -1)
14 # model forward
15 output_sequences = model.generate(
16 input_ids=prompt_tensor,
17 max_length=512,
18 top_p=0.95,
19 top_k=40,
20 temperature=1.0,
21 do_sample=True,
22 early_stopping=True,
23 bos_token_id=tokenizer.bos_token_id,
24 eos_token_id=tokenizer.eos_token_id,
25 pad_token_id=tokenizer.pad_token_id,
26 num_return_sequences=1
27 )
28
29 # convert model outputs to readable sentence
30 generated_sequence = output_sequences.tolist()[0]
31 generated_tokens = tokenizer.convert_ids_to_tokens(generated_sequence)
32 generated_text = tokenizer.convert_tokens_to_string(generated_tokens)
33 generated_text = "\n".join([s.strip() for s in generated_text.split('\\n')]).replace(' ', '\u3000').replace('<s>', '').replace('</s>', '\n\n---end---')
34 return generated_text
35
36
37print(gen_lyric("桜が咲く"))
38