nguyen-brat/zalo_cls

Model used

use to classification banner generate should has text or not used in zalo competition with banner task. i split train set with ratio 0.9, 0.1 for training and valid and achieve 0.693431 accuracy on valid test

used example

from transformers import AutoModelForSequenceClassification, AutoTokenizer
from torch.nn.functional import softmax
tokenizer = AutoTokenizer.from_pretrained('nguyen-brat/zalo_cls')
model_infer = AutoModelForSequenceClassification.from_pretrained('nguyen-brat/zalo_cls')
def inference(data_input):
  data_input = f' {tokenizer.sep_token} '.join(data_input)
  token = tokenizer(data_input, return_tensors='pt', padding=True, truncation=True)
  output = model_infer(**token)
  return softmax(output, dim=-1)