ko-sroberta-nli

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.

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 = ["์•ˆ๋…•ํ•˜์„ธ์š”?", "ํ•œ๊ตญ์–ด ๋ฌธ์žฅ ์ž„๋ฒ ๋”ฉ์„ ์œ„ํ•œ ๋ฒ„ํŠธ ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค."]

model = SentenceTransformer('jhgan/ko-sroberta-nli')
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 = ['This is an example sentence', 'Each sentence is converted']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('jhgan/ko-sroberta-nli')
model = AutoModel.from_pretrained('jhgan/ko-sroberta-nli')

# 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

KorNLI ํ•™์Šต ๋ฐ์ดํ„ฐ์…‹์œผ๋กœ ํ•™์Šตํ•œ ํ›„ KorSTS ํ‰๊ฐ€ ๋ฐ์ดํ„ฐ์…‹์œผ๋กœ ํ‰๊ฐ€ํ•œ ๊ฒฐ๊ณผ์ž…๋‹ˆ๋‹ค.

  • Cosine Pearson: 82.83
  • Cosine Spearman: 83.85
  • Euclidean Pearson: 82.87
  • Euclidean Spearman: 83.29
  • Manhattan Pearson: 82.88
  • Manhattan Spearman: 83.28
  • Dot Pearson: 80.34
  • Dot Spearman: 79.69

Training

The model was trained with the parameters:

DataLoader:

sentence_transformers.datasets.NoDuplicatesDataLoader.NoDuplicatesDataLoader of length 8885 with parameters:

{'batch_size': 64}

Loss:

sentence_transformers.losses.MultipleNegativesRankingLoss.MultipleNegativesRankingLoss with parameters:

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

Parameters of the fit()-Method:

{
    "epochs": 1,
    "evaluation_steps": 1000,
    "evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "<class 'transformers.optimization.AdamW'>",
    "optimizer_params": {
        "lr": 2e-05
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": null,
    "warmup_steps": 889,
    "weight_decay": 0.01
}

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: RobertaModel 
  (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})
)

Citing & Authors

  • Ham, J., Choe, Y. J., Park, K., Choi, I., & Soh, H. (2020). Kornli and korsts: New benchmark datasets for korean natural language understanding. arXiv preprint arXiv:2004.03289
  • Reimers, Nils and Iryna Gurevych. โ€œSentence-BERT: Sentence Embeddings using Siamese BERT-Networks.โ€ ArXiv abs/1908.10084 (2019)
  • Reimers, Nils and Iryna Gurevych. โ€œMaking Monolingual Sentence Embeddings Multilingual Using Knowledge Distillation.โ€ EMNLP (2020).
Downloads last month
6,110
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.