Views
No views yet
1import torch
2from transformers import T5Tokenizer, GPT2LMHeadModel
3
4device = torch.device("cpu")
5if torch.cuda.is_available():
6 device = torch.device("cuda")
7
8tokenizer = T5Tokenizer.from_pretrained("skytnt/gpt2-japanese-lyric-medium")
9model = GPT2LMHeadModel.from_pretrained("skytnt/gpt2-japanese-lyric-medium")
10model = model.to(device)
11
12def gen_lyric(title: str, prompt_text: str):
13 if len(title)!= 0 or len(prompt_text)!= 0:
14 prompt_text = "<s>" + title + "[CLS]" + prompt_text
15 prompt_text = prompt_text.replace("\n", "\\n ")
16 prompt_tokens = tokenizer.tokenize(prompt_text)
17 prompt_token_ids = tokenizer.convert_tokens_to_ids(prompt_tokens)
18 prompt_tensor = torch.LongTensor(prompt_token_ids)
19 prompt_tensor = prompt_tensor.view(1, -1).to(device)
20 else:
21 prompt_tensor = None
22 # model forward
23 output_sequences = model.generate(
24 input_ids=prompt_tensor,
25 max_length=512,
26 top_p=0.95,
27 top_k=40,
28 temperature=1.0,
29 do_sample=True,
30 early_stopping=True,
31 bos_token_id=tokenizer.bos_token_id,
32 eos_token_id=tokenizer.eos_token_id,
33 pad_token_id=tokenizer.pad_token_id,
34 num_return_sequences=1
35 )
36
37 # convert model outputs to readable sentence
38 generated_sequence = output_sequences.tolist()[0]
39 generated_tokens = tokenizer.convert_ids_to_tokens(generated_sequence)
40 generated_text = tokenizer.convert_tokens_to_string(generated_tokens)
41 generated_text = "\n".join([s.strip() for s in generated_text.split('\\n')]).replace(' ', '\u3000').replace('<s>', '').replace('</s>', '\n\n---end---')
42 title_and_lyric = generated_text.split("[CLS]",1)
43 if len(title_and_lyric)==1:
44 title,lyric = "" , title_and_lyric[0].strip()
45 else:
46 title,lyric = title_and_lyric[0].strip(), title_and_lyric[1].strip()
47 return f"---{title}---\n\n{lyric}"
48
49
50print(gen_lyric("桜",""))
51