kf-deberta-multitask

This is a sentence-transformers model: It maps sentences & paragraphs to a 768 dimensional dense vector space and can be used for tasks like clustering or semantic search. You can check the training recipes on GitHub.

Usage (Sentence-Transformers)

Using this model becomes easy when you have sentence-transformers installed:

pip install -U sentence-transformers

Then you can use the model like this:

from sentence_transformers import SentenceTransformer
sentences = ["안녕하세요?", "한국어 문장 임베딩을 위한 버트 모델입니다."]

model = SentenceTransformer("upskyy/kf-deberta-multitask")
embeddings = model.encode(sentences)
print(embeddings)

Usage (HuggingFace Transformers)

Without sentence-transformers, you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling-operation on-top of the contextualized word embeddings.

from transformers import AutoTokenizer, AutoModel
import torch


# Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0] # First element of model_output contains all token embeddings
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)


# Sentences we want sentence embeddings for
sentences = ["안녕하세요?", "한국어 문장 임베딩을 위한 버트 모델입니다."]

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained("upskyy/kf-deberta-multitask")
model = AutoModel.from_pretrained("upskyy/kf-deberta-multitask")

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)

# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Sentence embeddings:")
print(sentence_embeddings)

Evaluation Results

KorSTS, KorNLI 학습 데이터셋으로 멀티 태스크 학습을 진행한 후 KorSTS 평가 데이터셋으로 평가한 결과입니다.

• Cosine Pearson: 85.75
• Cosine Spearman: 86.25
• Manhattan Pearson: 84.80
• Manhattan Spearman: 85.27
• Euclidean Pearson: 84.79
• Euclidean Spearman: 85.25
• Dot Pearson: 82.93
• Dot Spearman: 82.86

model	cosine_pearson	cosine_spearman	euclidean_pearson	euclidean_spearman	manhattan_pearson	manhattan_spearman	dot_pearson	dot_spearman
kf-deberta-multitask	85.75	86.25	84.79	85.25	84.80	85.27	82.93	82.86
ko-sroberta-multitask	84.77	85.6	83.71	84.40	83.70	84.38	82.42	82.33
ko-sbert-multitask	84.13	84.71	82.42	82.66	82.41	82.69	80.05	79.69
ko-sroberta-base-nli	82.83	83.85	82.87	83.29	82.88	83.28	80.34	79.69
ko-sbert-nli	82.24	83.16	82.19	82.31	82.18	82.3	79.3	78.78
ko-sroberta-sts	81.84	81.82	81.15	81.25	81.14	81.25	79.09	78.54
ko-sbert-sts	81.55	81.23	79.94	79.79	79.9	79.75	76.02	75.31

Training

The model was trained with the parameters:

DataLoader:

sentence_transformers.datasets.NoDuplicatesDataLoader.NoDuplicatesDataLoader of length 4442 with parameters:

{'batch_size': 128}

Loss:

sentence_transformers.losses.MultipleNegativesRankingLoss.MultipleNegativesRankingLoss with parameters:

{'scale': 20.0, 'similarity_fct': 'cos_sim'}

DataLoader:

torch.utils.data.dataloader.DataLoader of length 719 with parameters:

{'batch_size': 8, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}

Loss:

sentence_transformers.losses.CosineSimilarityLoss.CosineSimilarityLoss

Parameters of the fit()-Method:

{
    "epochs": 10,
    "evaluation_steps": 1000,
    "evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
    "optimizer_params": {
        "lr": 2e-05
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": null,
    "warmup_steps": 719,
    "weight_decay": 0.01
}

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: DebertaV2Model 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False})
)

Citing & Authors

@proceedings{jeon-etal-2023-kfdeberta,
  title         = {KF-DeBERTa: Financial Domain-specific Pre-trained Language Model},
  author        = {Eunkwang Jeon, Jungdae Kim, Minsang Song, and Joohyun Ryu},
  booktitle     = {Proceedings of the 35th Annual Conference on Human and Cognitive Language Technology},
  moth          = {oct},
  year          = {2023},
  publisher     = {Korean Institute of Information Scientists and Engineers},
  url           = {http://www.hclt.kr/symp/?lnb=conference},
  pages         = {143--148},
}

@article{ham2020kornli,
  title={KorNLI and KorSTS: New Benchmark Datasets for Korean Natural Language Understanding},
  author={Ham, Jiyeon and Choe, Yo Joong and Park, Kyubyong and Choi, Ilji and Soh, Hyungjoon},
  journal={arXiv preprint arXiv:2004.03289},
  year={2020}
}