1from transformers import T5Tokenizer, T5ForConditionalGeneration
2
3model_name = "AtefAndrus/jmoji-t5-v4_focal_top50_20251224"
4tokenizer = T5Tokenizer.from_pretrained(model_name)
5model = T5ForConditionalGeneration.from_pretrained(model_name)
6
7text = "今日は楽しかった"
8inputs = tokenizer(text, return_tensors="pt", max_length=128, truncation=True)
9
10outputs = model.generate(
11 **inputs,
12 max_length=32,
13 do_sample=True,
14 temperature=0.7,
15 top_k=50,
16 top_p=0.95,
17 repetition_penalty=1.2, # 推奨: 過剰生成抑制
18)
19
20result = tokenizer.decode(outputs[0], skip_special_tokens=True)
21print(result) # 例: 😊 🎉 ✨
1@misc{jmoji2025,
2 author = {AtefAndrus},
3 title = {Jmoji: Japanese Text to Emoji Translation using Knowledge Distillation},
4 year = {2025},
5 publisher = {Hugging Face},
6 url = {https://huggingface.co/AtefAndrus/jmoji-t5-v4_focal_top50_20251224}
7}
