You can deploy this model and start earning money today!

mini1013/
master_item_top_bt3

Name: master_item_top_bt3
Availability: InStock

setfit

safetensors

roberta

sentence-transformers

text-classification

generated_from_setfit_trainer

Views

No views yet

SetFit with klue/roberta-base

This is a SetFit model that can be used for Text Classification. This SetFit model uses klue/roberta-base as the Sentence Transformer embedding model. A LogisticRegression instance is used for classification.

The model has been trained using an efficient few-shot learning technique that involves:

Fine-tuning a Sentence Transformer with contrastive learning.
Training a classification head with features from the fine-tuned Sentence Transformer.

Model Details

Model Description

Model Type: SetFit
Sentence Transformer body: klue/roberta-base
Classification head: a LogisticRegression instance
Maximum Sequence Length: 512 tokens
Number of Classes: 4 classes

Model Sources

Repository: SetFit on GitHub
Paper: Efficient Few-Shot Learning Without Prompts
Blogpost: SetFit: Efficient Few-Shot Learning Without Prompts

Model Labels

Label	Examples
3	'차앤박 CNP 안티포어 블랙헤드 클리어 키트 스트립 3세트(3회분) (#M)위메프 > 뷰티 > 스킨케어 > 팩/마스크 > 코팩 위메프 > 뷰티 > 스킨케어 > 팩/마스크 > 코팩' '미팩토리 3단 돼지코팩 10개입 × 3개 (#M)쿠팡 홈>뷰티>스킨케어>마스크/팩>패치/코팩>코팩 Coupang > 뷰티 > 스킨케어 > 마스크/팩' '[차앤박] CNP 안티포어 블랙헤드 버블 코팩 1매 / 넓은 모공 피부 / (#M)화장품/미용>마스크/팩>코팩 Naverstore > 화장품/미용 > 마스크/팩 > 코팩'
0	'메디힐×마리끌레르 기획전 앰플/크림/마스크팩~58% 25_메디힐 티트리 케어솔루션 에센셜마스크 [10매] 쇼킹딜 홈>뷰티>클렌징/팩/마스크>팩/마스크;11st>스킨케어>팩/마스크>마스크시트팩;(#M)11st>뷰티>클렌징/팩/마스크>팩/마스크 11st Hour Event > 패션/뷰티 > 뷰티 > 클렌징/팩/마스크 > 팩/마스크' '[의료기기] 듀오덤 스팟패치 72매 [의료기기] 듀오덤 스팟패치 72매 (#M)홈>구강/건강용품>패치/겔>스팟패치 OLIVEYOUNG > 베스트 > 구강/건강용품' '이지덤 뷰티 릴리프 스팟패치 57개입 3개 (#M)쿠팡 홈>생활용품>건강/의료용품>의약외품/상비용품>반창고/밴드 Coupang > 뷰티 > 스킨케어 > 마스크/팩 > 패치/코팩 > 스팟패치'
2	'안스킨 클래리파잉 골드 모델링 팩 1000ml 20개 (#M)홈>화장품/미용>마스크/팩>필오프팩 Naverstore > 화장품/미용 > 마스크/팩 > 필오프팩' '[러쉬]오티픽스 75g - 프레쉬 페이스 마스크/마스크 팩 ssg > 뷰티 > 스킨케어 > 마스크/팩 > 시트마스크;ssg > 뷰티 > 헤어/바디 > 세정/입욕용품 > 입욕제/버블바스;ssg > 뷰티 > 스킨케어 > 마스크/팩;ssg > 뷰티 > 스킨케어 > 클렌징 ssg > 뷰티 > 스킨케어 > 마스크/팩 > 시트마스크' '푸드어홀릭 콜라겐 필오프팩 150ml / 다시마 MinSellAmount (#M)화장품/향수>팩/마스크>필오프팩 Gmarket > 뷰티 > 화장품/향수 > 팩/마스크 > 필오프팩'
1	'물광 콜라겐 크림 티르티르 80ml 생크림 도자기 피부 물광마스크 이유빈 콜라겐물광마스크40ml (#M)홈>전체상품 Naverstore > 화장품/미용 > 남성화장품 > 크림' '립 슬리핑 마스크 EX 20g 4종 베리 자몽 민트초코 애플라임 베리 (#M)홈>화장품/미용>마스크/팩>수면팩 Naverstore > 화장품/미용 > 마스크/팩 > 수면팩' '설화수 한방 슬리핑마스크 나이트여운팩 120ml 1개 (#M)위메프 > 뷰티 > 스킨케어 > 팩/마스크 > 수면팩 위메프 > 뷰티 > 스킨케어 > 팩/마스크 > 수면팩'

Evaluation

Metrics

Label	Accuracy
all	0.7775

Uses

Direct Use for Inference

First install the SetFit library:

pip install setfit

Then you can load this model and run inference.

python

from setfit import SetFitModel

# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("mini1013/master_item_top_bt3")
# Run inference
preds = model("[대용량] 라네즈 크림 스킨 퀵 스킨 팩 100매(140ml) 피부진정 보습  (#M)홈>라네즈 Naverstore > 화장품/미용 > 마스크/팩 > 수면팩")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	11	21.75	91

Label	Training Sample Count
0	50
1	50
2	50
3	50

Training Hyperparameters

batch_size: (64, 64)
num_epochs: (30, 30)
max_steps: -1
sampling_strategy: oversampling
num_iterations: 100
body_learning_rate: (2e-05, 1e-05)
head_learning_rate: 0.01
loss: CosineSimilarityLoss
distance_metric: cosine_distance
margin: 0.25
end_to_end: False
use_amp: False
warmup_proportion: 0.1
l2_weight: 0.01
seed: 42
eval_max_steps: -1
load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0032	1	0.4549	-
0.1597	50	0.3933	-
0.3195	100	0.3669	-
0.4792	150	0.2841	-
0.6390	200	0.1163	-
0.7987	250	0.0104	-
0.9585	300	0.0072	-
1.1182	350	0.0065	-
1.2780	400	0.0059	-
1.4377	450	0.0058	-
1.5974	500	0.0035	-
1.7572	550	0.0032	-
1.9169	600	0.0032	-
2.0767	650	0.0025	-
2.2364	700	0.0023	-
2.3962	750	0.0023	-
2.5559	800	0.0025	-
2.7157	850	0.0023	-
2.8754	900	0.003	-
3.0351	950	0.0026	-
3.1949	1000	0.0043	-
3.3546	1050	0.0022	-
3.5144	1100	0.0024	-
3.6741	1150	0.0025	-
3.8339	1200	0.0025	-
3.9936	1250	0.0024	-
4.1534	1300	0.0025	-
4.3131	1350	0.0025	-
4.4728	1400	0.0027	-
4.6326	1450	0.0023	-
4.7923	1500	0.0022	-
4.9521	1550	0.0026	-
5.1118	1600	0.0022	-
5.2716	1650	0.0027	-
5.4313	1700	0.0022	-
5.5911	1750	0.0024	-
5.7508	1800	0.0029	-
5.9105	1850	0.0018	-
6.0703	1900	0.0033	-
6.2300	1950	0.002	-
6.3898	2000	0.0027	-
6.5495	2050	0.0021	-
6.7093	2100	0.0022	-
6.8690	2150	0.0023	-
7.0288	2200	0.0026	-
7.1885	2250	0.0018	-
7.3482	2300	0.0024	-
7.5080	2350	0.002	-
7.6677	2400	0.0027	-
7.8275	2450	0.0022	-
7.9872	2500	0.0032	-
8.1470	2550	0.0029	-
8.3067	2600	0.0025	-
8.4665	2650	0.0017	-
8.6262	2700	0.0026	-
8.7859	2750	0.0023	-
8.9457	2800	0.0023	-
9.1054	2850	0.0029	-
9.2652	2900	0.0028	-
9.4249	2950	0.0021	-
9.5847	3000	0.0027	-
9.7444	3050	0.0019	-
9.9042	3100	0.0022	-
10.0639	3150	0.003	-
10.2236	3200	0.0024	-
10.3834	3250	0.0019	-
10.5431	3300	0.0023	-
10.7029	3350	0.0024	-
10.8626	3400	0.0026	-
11.0224	3450	0.0025	-
11.1821	3500	0.0022	-
11.3419	3550	0.0023	-
11.5016	3600	0.0027	-
11.6613	3650	0.0032	-
11.8211	3700	0.0022	-
11.9808	3750	0.0019	-
12.1406	3800	0.0029	-
12.3003	3850	0.0026	-
12.4601	3900	0.0027	-
12.6198	3950	0.0019	-
12.7796	4000	0.0021	-
12.9393	4050	0.0023	-
13.0990	4100	0.0027	-
13.2588	4150	0.0021	-
13.4185	4200	0.0022	-
13.5783	4250	0.0026	-
13.7380	4300	0.0025	-
13.8978	4350	0.0025	-
14.0575	4400	0.0021	-
14.2173	4450	0.0031	-
14.3770	4500	0.0022	-
14.5367	4550	0.0016	-
14.6965	4600	0.0027	-
14.8562	4650	0.0027	-
15.0160	4700	0.0027	-
15.1757	4750	0.0021	-
15.3355	4800	0.0027	-
15.4952	4850	0.0031	-
15.6550	4900	0.0021	-
15.8147	4950	0.0023	-
15.9744	5000	0.002	-
16.1342	5050	0.0024	-
16.2939	5100	0.0026	-
16.4537	5150	0.002	-
16.6134	5200	0.0026	-
16.7732	5250	0.0029	-
16.9329	5300	0.0023	-
17.0927	5350	0.0022	-
17.2524	5400	0.0028	-
17.4121	5450	0.0026	-
17.5719	5500	0.0017	-
17.7316	5550	0.0032	-
17.8914	5600	0.0022	-
18.0511	5650	0.0019	-
18.2109	5700	0.0024	-
18.3706	5750	0.0026	-
18.5304	5800	0.0031	-
18.6901	5850	0.0024	-
18.8498	5900	0.0018	-
19.0096	5950	0.0023	-
19.1693	6000	0.0025	-
19.3291	6050	0.0028	-
19.4888	6100	0.002	-
19.6486	6150	0.0026	-
19.8083	6200	0.0022	-
19.9681	6250	0.0025	-
20.1278	6300	0.0022	-
20.2875	6350	0.0025	-
20.4473	6400	0.0024	-
20.6070	6450	0.0027	-
20.7668	6500	0.0017	-
20.9265	6550	0.0025	-
21.0863	6600	0.0025	-
21.2460	6650	0.002	-
21.4058	6700	0.0033	-
21.5655	6750	0.0021	-
21.7252	6800	0.0022	-
21.8850	6850	0.0027	-
22.0447	6900	0.0021	-
22.2045	6950	0.0028	-
22.3642	7000	0.0021	-
22.5240	7050	0.0021	-
22.6837	7100	0.0027	-
22.8435	7150	0.0021	-
23.0032	7200	0.0029	-
23.1629	7250	0.0036	-
23.3227	7300	0.002	-
23.4824	7350	0.0021	-
23.6422	7400	0.002	-
23.8019	7450	0.0025	-
23.9617	7500	0.0024	-
24.1214	7550	0.0026	-
24.2812	7600	0.002	-
24.4409	7650	0.0024	-
24.6006	7700	0.0025	-
24.7604	7750	0.0023	-
24.9201	7800	0.0027	-
25.0799	7850	0.0023	-
25.2396	7900	0.0024	-
25.3994	7950	0.0027	-
25.5591	8000	0.0038	-
25.7188	8050	0.0065	-
25.8786	8100	0.0037	-
26.0383	8150	0.0032	-
26.1981	8200	0.0031	-
26.3578	8250	0.0028	-
26.5176	8300	0.0024	-
26.6773	8350	0.0023	-
26.8371	8400	0.0028	-
26.9968	8450	0.0023	-
27.1565	8500	0.0028	-
27.3163	8550	0.0025	-
27.4760	8600	0.0027	-
27.6358	8650	0.002	-
27.7955	8700	0.0024	-
27.9553	8750	0.0023	-
28.1150	8800	0.0029	-
28.2748	8850	0.0025	-
28.4345	8900	0.002	-
28.5942	8950	0.0025	-
28.7540	9000	0.002	-
28.9137	9050	0.0027	-
29.0735	9100	0.0028	-
29.2332	9150	0.0016	-
29.3930	9200	0.0032	-
29.5527	9250	0.0026	-
29.7125	9300	0.0025	-
29.8722	9350	0.0025	-

Framework Versions

Python: 3.10.12
SetFit: 1.1.0
Sentence Transformers: 3.3.1
Transformers: 4.44.2
PyTorch: 2.2.0a0+81ea7a4
Datasets: 3.2.0
Tokenizers: 0.19.1

Citation

BibTeX

bibtex

@article{https://doi.org/10.48550/arxiv.2209.11055,
    doi = {10.48550/ARXIV.2209.11055},
    url = {https://arxiv.org/abs/2209.11055},
    author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
    keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
    title = {Efficient Few-Shot Learning Without Prompts},
    publisher = {arXiv},
    year = {2022},
    copyright = {Creative Commons Attribution 4.0 International}
}