klue-tc-bert-base-multilingual-cased

Model information

• language : Korean
• fine tuning data : klue-tc (a.k.a. YNAT)
• License : CC-BY-SA 4.0
• Base model : bert-base-multilingual-cased
• input : news headline
• output : topic

---

Train information

• train_runtime: 1477.3876
• train_steps_per_second: 2.416
• train_loss: 0.3722160959110207
• epoch: 5.0

---

How to use

from transformers import AutoTokenizer, AutoModelForSequenceClassification
  
tokenizer = AutoTokenizer.from_pretrained (
"seongju/klue-tc-bert-base-multilingual-cased"
)

model = AutoModelForSequenceClassification.from_pretrained (
"seongju/klue-tc-bert-base-multilingual-cased"
)
mapping = {0: 'IT과학', 1: '경제', 2: '사회', 
3: '생활문화', 4: '세계', 5: '스포츠', 6: '정치'}
inputs = tokenizer(
"백신 회피 가능성? 남미에서 새로운 변이 바이러스 급속 확산 ",
 padding=True, truncation=True, max_length=128, return_tensors="pt"
 )
outputs = model(**inputs)
probs = outputs[0].softmax(1)
output = mapping[probs.argmax().item()]