Update README.md
Browse files
README.md
CHANGED
@@ -5,14 +5,27 @@ tags:
|
|
5 |
- feature-extraction
|
6 |
- sentence-similarity
|
7 |
- transformers
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
---
|
10 |
|
11 |
# smartmind/roberta-ko-small-tsdae
|
12 |
|
13 |
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 256 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
14 |
|
15 |
-
|
|
|
|
|
|
|
|
|
16 |
|
17 |
## Usage (Sentence-Transformers)
|
18 |
|
@@ -72,16 +85,18 @@ print(sentence_embeddings)
|
|
72 |
|
73 |
## Evaluation Results
|
74 |
|
75 |
-
|
76 |
-
|
77 |
-
For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name=smartmind/roberta-ko-small-tsdae)
|
78 |
|
|
|
|
|
|
|
|
|
79 |
|
80 |
|
81 |
## Full Model Architecture
|
82 |
```
|
83 |
SentenceTransformer(
|
84 |
-
(0): Transformer({'max_seq_length': 508, 'do_lower_case': False}) with Transformer model: RobertaModel
|
85 |
(1): Pooling({'word_embedding_dimension': 256, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
|
86 |
)
|
87 |
```
|
|
|
5 |
- feature-extraction
|
6 |
- sentence-similarity
|
7 |
- transformers
|
8 |
+
language:
|
9 |
+
- ko
|
10 |
+
license:
|
11 |
+
- mit
|
12 |
+
widget:
|
13 |
+
source_sentence: "대한민국의 수도는 서울입니다."
|
14 |
+
sentences:
|
15 |
+
- "미국의 수도는 뉴욕이 아닙니다."
|
16 |
+
- "대한민국의 수도 요금은 저렴한 편입니다."
|
17 |
+
- "서울은 대한민국의 수도입니다."
|
18 |
---
|
19 |
|
20 |
# smartmind/roberta-ko-small-tsdae
|
21 |
|
22 |
This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 256 dimensional dense vector space and can be used for tasks like clustering or semantic search.
|
23 |
|
24 |
+
Korean roberta small model pretrained with [TSDAE](https://arxiv.org/abs/2104.06979).
|
25 |
+
|
26 |
+
[TSDAE](https://arxiv.org/abs/2104.06979)로 사전학습된 한국어 roberta모델입니다. 모델의 구조는 [lassl/roberta-ko-small](https://huggingface.co/lassl/roberta-ko-small)과 동일합니다. 토크나이저는 다릅니다.
|
27 |
+
|
28 |
+
sentence-similarity를 구하는 용도로 바로 사용할 수도 있고, 목적에 맞게 파인튜닝하여 사용할 수도 있습니다.
|
29 |
|
30 |
## Usage (Sentence-Transformers)
|
31 |
|
|
|
85 |
|
86 |
## Evaluation Results
|
87 |
|
88 |
+
[klue](https://huggingface.co/datasets/klue) STS 데이터에 대해 다음 점수를 얻었습니다. 이 데이터에 대해 파인튜닝하지 **않은** 상태로 구한 점수입니다.
|
|
|
|
|
89 |
|
90 |
+
|split|cosine_pearson|cosine_spearman|euclidean_pearson|euclidean_spearman|manhattan_pearson|manhattan_spearman|dot_pearson|dot_spearman|
|
91 |
+
|-----|--------------|---------------|-----------------|------------------|-----------------|------------------|-----------|------------|
|
92 |
+
|train|0.8735|0.8676|0.8268|0.8357|0.8248|0.8336|0.8449|0.8383|
|
93 |
+
|validation|0.5409|0.5349|0.4786|0.4657|0.4775|0.4625|0.5284|0.5252|
|
94 |
|
95 |
|
96 |
## Full Model Architecture
|
97 |
```
|
98 |
SentenceTransformer(
|
99 |
+
(0): Transformer({'max_seq_length': 508, 'do_lower_case': False}) with Transformer model: RobertaModel
|
100 |
(1): Pooling({'word_embedding_dimension': 256, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
|
101 |
)
|
102 |
```
|