klue-sroberta-continue-learning-by-mnr

해당 모델은 KLUE/NLI, KLUE/STS 데이터셋을 활용하였으며, sentence-transformers의 공식 문서 내 소개된 continue-learning 방법을 통해 아래와 같이 학습되었습니다.

1. NLI 데이터셋을 통해 nagative sampling 후, MultipleNegativeRankingLoss를 활용하여 1차 NLI training 수행
2. 1에서 학습완료 된 모델에 STS 데이터셋을 통해, CosineSimilarityLoss를 활용하여 2차 STS training 수행

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 = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer("bespin-global/klue-sroberta-base-continue-learning-by-mnr")
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("bespin-global/klue-sroberta-base-continue-learning-by-mnr")
model = AutoModel.from_pretrained("bespin-global/klue-sroberta-base-continue-learning-by-mnr")

# 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

EmbeddingSimilarityEvaluator: Evaluating the model on sts-test dataset:

• Cosine-Similarity :
  • Pearson: 0.8901 Spearman: 0.8893
• Manhattan-Distance:
  • Pearson: 0.8867 Spearman: 0.8818
• Euclidean-Distance:
  • Pearson: 0.8875 Spearman: 0.8827
• Dot-Product-Similarity:
  • Pearson: 0.8786 Spearman: 0.8735
• Average : 0.8892573547643868

Training

The model was trained with the parameters:

DataLoader:

torch.utils.data.dataloader.DataLoader of length 329 with parameters:

{'batch_size': 32, '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": 4,
    "evaluation_steps": 32,
    "evaluator": "sentence_transformers.evaluation.EmbeddingSimilarityEvaluator.EmbeddingSimilarityEvaluator",
    "max_grad_norm": 1,
    "optimizer_class": "",
    "optimizer_params": {
        "lr": 2e-05
    },
    "scheduler": "WarmupLinear",
    "steps_per_epoch": null,
    "warmup_steps": 132,
    "weight_decay": 0.01
}

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) 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

JaeHyeong AN at Bespin Global