mini1013/master_item_top_bt6

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:

1. Fine-tuning a Sentence Transformer with contrastive learning.
2. 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: 8 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
7	'[JAJU/자주] 원형 리필 공병 통 110ml ssg > 뷰티 > 미용기기/소품 > 거울/용기/기타소품;ssg > 뷰티 > 헤어/바디/미용/구강 > 미용기기 ssg > 뷰티 > 미용기기/소품 > 거울/용기/기타소품' '세맘스 아기랑 + 엄마랑 파우치 세트 핑크스마일_엄마(가로 11.5cm x 세로 13cm), 아기(가로 8cm x 세로 10.5cm) (#M)쿠팡 홈>여행용품>여행파우치>화장품파우치 Coupang > 뷰티 > 뷰티소품 > 용기/거울/기타소품 > 파우치' '라인 프린팅 파스텔컬러 롤온공병 10ml 6종 세트 흰색(뚜껑) × 1세트 (#M)쿠팡 홈>뷰티>뷰티소품>용기/거울/기타소품>기타소품 Coupang > 뷰티 > 뷰티소품 > 용기/거울/기타소품 > 기타소품'
3	'트위저맨 슬랜트 트위저 족집게 베이비 핑크 × 9개 (#M)쿠팡 홈>뷰티>뷰티소품>아이소품>족집게/샤프너 Coupang > 뷰티 > 뷰티소품 > 아이소품 > 족집게/샤프너' '트위저맨 미니 슬랜트 트위저 로즈골드 265161 LotteOn > 뷰티 > 메이크업 > 베이스메이크업 > 파운데이션 LotteOn > 뷰티 > 메이크업 > 베이스메이크업 > 파운데이션' '트위저맨 클래식 슬랜트 트위저 베이비핑크, 1개 LotteOn > 뷰티 > 뷰티기기/소품 > 메이크업소품 > 브러쉬 LotteOn > 뷰티 > 뷰티기기/소품 > 메이크업소품 > 브러쉬'
6	'천일 매직 롯드 10P 1호~6호 뿌리볼륨롯드 파마롯드 매직롯드 5호_1개 홈>화장품/미용>뷰티소품>헤어소품>헤어롤;홈>전체상품;(#M)홈>롯드 Naverstore > 화장품/미용 > 뷰티소품 > 헤어소품 > 헤어롤' '다이슨 45mm 35mm 롤브러쉬 대왕롤빗 엉킴방지빗 니켈블랙 (#M)홈>미용건강 Naverstore > 화장품/미용 > 뷰티소품 > 헤어소품 > 헤어브러시' '프리시전 섀이더 브러쉬 스몰 단품없음 LotteOn > 뷰티 > 뷰티기기 > 액세서리/소모품 LotteOn > 뷰티 > 뷰티기기 > 액세서리/소모품'
0	'천연 자초 립밤 만들기 키트 diy 향 선택(8개) 사과+에탄올20ml (#M)홈>비누&립밤&세제 만들기>만들기키트 Naverstore > 화장품/미용 > 색조메이크업 > 립케어'
5	'프로 피니쉬 스폰지 단품없음 LotteOn > 뷰티 > 뷰티기기 > 액세서리/소모품 LotteOn > 뷰티 > 뷰티기기 > 액세서리/소모품' 'JAJU 사각 면봉_화장 겸용 200P 기타_FR LotteOn > 뷰티 > 뷰티기기/소품 > 위생용품 > 면봉 LotteOn > 뷰티 > 뷰티기기/소품 > 위생용품 > 면봉' 'mts 롤러 기계 MTS 스탬프 앰플 바르는 도구 니들 빠른흡수 상품선택_2-더마롤러-0.3mm LotteOn > 뷰티 > 뷰티기기/소품 > 피부케어기 > 피부케어기 LotteOn > 뷰티 > 뷰티기기/소품 > 피부케어기 > 피부케어기'
1	'더툴랩 101B 베이비태스커 파운데이션 베이스 메이크업 브러쉬 쿠션브러쉬 236097 (#M)홈>화장품/미용>뷰티소품>메이크업브러시>브러시세트 Naverstore > 화장품/미용 > 뷰티소품 > 메이크업브러시 > 브러시세트' '더툴랩 204 블렌딩 아이섀도우 스몰 총알 브러쉬 (#M)화장품/미용>뷰티소품>페이스소품>코털제거기 AD > Naverstore > 화장품/미용 > 뷰티소품 > 페이스소품 > 코털제거기' '더툴랩 브러쉬 231 컨실러 파운데이션 (#M)화장품/미용>뷰티소품>메이크업브러시>페이스브러시 LO > Naverstore > 화장품/미용 > 뷰티소품 > 메이크업브러시 > 페이스브러시'
2	'요들가운 미용실 LC 커트보 어깨보 컷트보 인쇄가능 15.모델210T커트보_블랙 (#M)홈>가운,유니폼>컷트보 Naverstore > 화장품/미용 > 뷰티소품 > 헤어소품 > 기타헤어소품' '요들가운 미용실 LC 커트보 어깨보 컷트보 인쇄가능 12.듀스포체크 커트보_퍼플 (#M)홈>가운,유니폼>컷트보 Naverstore > 화장품/미용 > 뷰티소품 > 헤어소품 > 기타헤어소품' '[백화점][JPClarisse] 장폴클라리쎄 거미 왕대 집게핀 JPSA0001 진베이지 (#M)GSSHOP>뷰티>뷰티소품>헤어소품 GSSHOP > 뷰티 > 뷰티소품 > 헤어소품 > 헤어집게'
4	'레터링 쇄골 현아 타투 스티커 30장 마스크 판박이 3타투세트30장-수채화 LotteOn > 뷰티 > 마스크/팩 > 기타패치 LotteOn > 뷰티 > 마스크/팩 > 기타패치' '산리오 캐릭터 타투 스티커 어린이 문신 마스크판박이 5.헬로키티(2매입) 홈>패션잡화🛍>잡화🐱\u200d💻;(#M)홈>캐릭터🙂>산리오 Naverstore > 화장품/미용 > 뷰티소품 > 타투' '문신 타투 스티커 바디 형 쇄골 반팔 레터링 흉터 커버__개성 다이소 헤나 다목적 노출 패션 미용 다용도 추천 패셔니스타 여름 A type 타투스티커 30종세트 (#M)SSG.COM/헤어/바디/슬리밍/푸드/기타용품/타투 ssg > 뷰티 > 헤어/바디 > 슬리밍/푸드/기타용품 > 타투'

Evaluation

Metrics

Label	Accuracy
all	0.9419

Uses

Direct Use for Inference

First install the SetFit library:

pip install setfit

Then you can load this model and run inference.

from setfit import SetFitModel

# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("mini1013/master_item_top_bt6")
# Run inference
preds = model("에스쁘아 에어 퍼프 5개입 소프트 터치 에어퍼프 LotteOn > 뷰티 > 뷰티기기/소품 > 메이크업소품 > 퍼프 LotteOn > 뷰티 > 뷰티기기/소품 > 메이크업소품 > 브러쉬")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	12	22.0313	72

Label	Training Sample Count
0	1
1	50
2	50
3	50
4	50
5	50
6	50
7	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
0.0018	1	0.4099	-
0.0911	50	0.3973	-
0.1821	100	0.3456	-
0.2732	150	0.2947	-
0.3643	200	0.2369	-
0.4554	250	0.1705	-
0.5464	300	0.107	-
0.6375	350	0.0696	-
0.7286	400	0.0494	-
0.8197	450	0.0488	-
0.9107	500	0.0307	-
1.0018	550	0.0259	-
1.0929	600	0.0247	-
1.1840	650	0.022	-
1.2750	700	0.0215	-
1.3661	750	0.005	-
1.4572	800	0.0007	-
1.5483	850	0.0004	-
1.6393	900	0.0002	-
1.7304	950	0.0001	-
1.8215	1000	0.0001	-
1.9126	1050	0.0001	-
2.0036	1100	0.0001	-
2.0947	1150	0.0001	-
2.1858	1200	0.0001	-
2.2769	1250	0.0	-
2.3679	1300	0.0	-
2.4590	1350	0.0	-
2.5501	1400	0.0	-
2.6412	1450	0.0	-
2.7322	1500	0.0	-
2.8233	1550	0.0	-
2.9144	1600	0.0	-
3.0055	1650	0.0	-
3.0965	1700	0.0	-
3.1876	1750	0.0	-
3.2787	1800	0.0	-
3.3698	1850	0.0	-
3.4608	1900	0.0	-
3.5519	1950	0.0	-
3.6430	2000	0.0	-
3.7341	2050	0.0	-
3.8251	2100	0.0	-
3.9162	2150	0.0	-
4.0073	2200	0.0	-
4.0984	2250	0.0	-
4.1894	2300	0.0	-
4.2805	2350	0.0	-
4.3716	2400	0.0	-
4.4627	2450	0.0	-
4.5537	2500	0.0	-
4.6448	2550	0.0	-
4.7359	2600	0.0	-
4.8270	2650	0.0	-
4.9180	2700	0.0	-
5.0091	2750	0.0	-
5.1002	2800	0.0	-
5.1913	2850	0.0	-
5.2823	2900	0.0	-
5.3734	2950	0.0	-
5.4645	3000	0.0	-
5.5556	3050	0.0	-
5.6466	3100	0.0	-
5.7377	3150	0.0	-
5.8288	3200	0.0	-
5.9199	3250	0.0	-
6.0109	3300	0.0	-
6.1020	3350	0.0	-
6.1931	3400	0.0	-
6.2842	3450	0.0	-
6.3752	3500	0.0	-
6.4663	3550	0.0	-
6.5574	3600	0.0	-
6.6485	3650	0.0	-
6.7395	3700	0.0	-
6.8306	3750	0.0	-
6.9217	3800	0.0	-
7.0128	3850	0.0	-
7.1038	3900	0.0	-
7.1949	3950	0.0	-
7.2860	4000	0.0	-
7.3770	4050	0.0	-
7.4681	4100	0.0	-
7.5592	4150	0.0	-
7.6503	4200	0.0	-
7.7413	4250	0.0	-
7.8324	4300	0.0	-
7.9235	4350	0.0	-
8.0146	4400	0.0	-
8.1056	4450	0.0	-
8.1967	4500	0.0	-
8.2878	4550	0.0	-
8.3789	4600	0.0	-
8.4699	4650	0.0	-
8.5610	4700	0.0	-
8.6521	4750	0.0	-
8.7432	4800	0.0	-
8.8342	4850	0.0	-
8.9253	4900	0.0	-
9.0164	4950	0.0	-
9.1075	5000	0.0	-
9.1985	5050	0.0	-
9.2896	5100	0.0	-
9.3807	5150	0.0	-
9.4718	5200	0.0	-
9.5628	5250	0.0	-
9.6539	5300	0.0	-
9.7450	5350	0.0	-
9.8361	5400	0.0	-
9.9271	5450	0.0	-
10.0182	5500	0.0	-
10.1093	5550	0.0	-
10.2004	5600	0.0	-
10.2914	5650	0.0	-
10.3825	5700	0.0	-
10.4736	5750	0.0	-
10.5647	5800	0.0	-
10.6557	5850	0.0	-
10.7468	5900	0.0	-
10.8379	5950	0.0	-
10.9290	6000	0.0	-
11.0200	6050	0.0	-
11.1111	6100	0.0	-
11.2022	6150	0.0	-
11.2933	6200	0.0	-
11.3843	6250	0.0	-
11.4754	6300	0.0	-
11.5665	6350	0.0	-
11.6576	6400	0.0	-
11.7486	6450	0.0	-
11.8397	6500	0.0	-
11.9308	6550	0.0	-
12.0219	6600	0.0	-
12.1129	6650	0.0	-
12.2040	6700	0.0	-
12.2951	6750	0.0	-
12.3862	6800	0.0	-
12.4772	6850	0.0	-
12.5683	6900	0.0	-
12.6594	6950	0.0	-
12.7505	7000	0.0	-
12.8415	7050	0.0	-
12.9326	7100	0.0	-
13.0237	7150	0.0	-
13.1148	7200	0.0	-
13.2058	7250	0.0	-
13.2969	7300	0.0	-
13.3880	7350	0.0	-
13.4791	7400	0.0	-
13.5701	7450	0.0	-
13.6612	7500	0.0	-
13.7523	7550	0.0	-
13.8434	7600	0.0	-
13.9344	7650	0.0	-
14.0255	7700	0.0	-
14.1166	7750	0.0	-
14.2077	7800	0.0	-
14.2987	7850	0.0	-
14.3898	7900	0.0	-
14.4809	7950	0.0	-
14.5719	8000	0.0	-
14.6630	8050	0.0	-
14.7541	8100	0.0	-
14.8452	8150	0.0	-
14.9362	8200	0.0	-
15.0273	8250	0.0	-
15.1184	8300	0.0	-
15.2095	8350	0.0	-
15.3005	8400	0.0	-
15.3916	8450	0.0	-
15.4827	8500	0.0	-
15.5738	8550	0.012	-
15.6648	8600	0.0012	-
15.7559	8650	0.0003	-
15.8470	8700	0.0	-
15.9381	8750	0.0	-
16.0291	8800	0.0	-
16.1202	8850	0.0	-
16.2113	8900	0.0	-
16.3024	8950	0.0	-
16.3934	9000	0.0	-
16.4845	9050	0.0	-
16.5756	9100	0.0	-
16.6667	9150	0.0	-
16.7577	9200	0.0	-
16.8488	9250	0.0	-
16.9399	9300	0.0	-
17.0310	9350	0.0	-
17.1220	9400	0.0	-
17.2131	9450	0.0	-
17.3042	9500	0.0	-
17.3953	9550	0.0	-
17.4863	9600	0.0	-
17.5774	9650	0.0	-
17.6685	9700	0.0	-
17.7596	9750	0.0	-
17.8506	9800	0.0	-
17.9417	9850	0.0	-
18.0328	9900	0.0	-
18.1239	9950	0.0	-
18.2149	10000	0.0	-
18.3060	10050	0.0	-
18.3971	10100	0.0	-
18.4882	10150	0.0	-
18.5792	10200	0.0	-
18.6703	10250	0.0	-
18.7614	10300	0.0	-
18.8525	10350	0.0	-
18.9435	10400	0.0	-
19.0346	10450	0.0	-
19.1257	10500	0.0	-
19.2168	10550	0.0	-
19.3078	10600	0.0	-
19.3989	10650	0.0	-
19.4900	10700	0.0	-
19.5811	10750	0.0	-
19.6721	10800	0.0	-
19.7632	10850	0.0	-
19.8543	10900	0.0	-
19.9454	10950	0.0	-
20.0364	11000	0.0	-
20.1275	11050	0.0	-
20.2186	11100	0.0	-
20.3097	11150	0.0	-
20.4007	11200	0.0	-
20.4918	11250	0.0	-
20.5829	11300	0.0	-
20.6740	11350	0.0	-
20.7650	11400	0.0	-
20.8561	11450	0.0	-
20.9472	11500	0.0	-
21.0383	11550	0.0	-
21.1293	11600	0.0	-
21.2204	11650	0.0	-
21.3115	11700	0.0	-
21.4026	11750	0.0	-
21.4936	11800	0.0	-
21.5847	11850	0.0	-
21.6758	11900	0.0	-
21.7668	11950	0.0	-
21.8579	12000	0.0	-
21.9490	12050	0.0	-
22.0401	12100	0.0	-
22.1311	12150	0.0	-
22.2222	12200	0.0	-
22.3133	12250	0.0	-
22.4044	12300	0.0	-
22.4954	12350	0.0	-
22.5865	12400	0.0	-
22.6776	12450	0.0	-
22.7687	12500	0.0	-
22.8597	12550	0.0	-
22.9508	12600	0.0	-
23.0419	12650	0.0	-
23.1330	12700	0.0	-
23.2240	12750	0.0	-
23.3151	12800	0.0	-
23.4062	12850	0.0	-
23.4973	12900	0.0	-
23.5883	12950	0.0	-
23.6794	13000	0.0	-
23.7705	13050	0.0	-
23.8616	13100	0.0	-
23.9526	13150	0.0	-
24.0437	13200	0.0	-
24.1348	13250	0.0	-
24.2259	13300	0.0	-
24.3169	13350	0.0	-
24.4080	13400	0.0	-
24.4991	13450	0.0	-
24.5902	13500	0.0	-
24.6812	13550	0.0	-
24.7723	13600	0.0	-
24.8634	13650	0.0	-
24.9545	13700	0.0	-
25.0455	13750	0.0	-
25.1366	13800	0.0	-
25.2277	13850	0.0	-
25.3188	13900	0.0	-
25.4098	13950	0.0	-
25.5009	14000	0.0	-
25.5920	14050	0.0	-
25.6831	14100	0.0	-
25.7741	14150	0.0	-
25.8652	14200	0.0	-
25.9563	14250	0.0	-
26.0474	14300	0.0	-
26.1384	14350	0.0	-
26.2295	14400	0.0	-
26.3206	14450	0.0	-
26.4117	14500	0.0	-
26.5027	14550	0.0	-
26.5938	14600	0.0	-
26.6849	14650	0.0	-
26.7760	14700	0.0	-
26.8670	14750	0.0	-
26.9581	14800	0.0	-
27.0492	14850	0.0	-
27.1403	14900	0.0	-
27.2313	14950	0.0	-
27.3224	15000	0.0	-
27.4135	15050	0.0	-
27.5046	15100	0.0	-
27.5956	15150	0.0	-
27.6867	15200	0.0	-
27.7778	15250	0.0	-
27.8689	15300	0.0	-
27.9599	15350	0.0	-
28.0510	15400	0.0	-
28.1421	15450	0.0	-
28.2332	15500	0.0	-
28.3242	15550	0.0	-
28.4153	15600	0.0	-
28.5064	15650	0.0	-
28.5974	15700	0.0	-
28.6885	15750	0.0	-
28.7796	15800	0.0	-
28.8707	15850	0.0	-
28.9617	15900	0.0	-
29.0528	15950	0.0	-
29.1439	16000	0.0	-
29.2350	16050	0.0	-
29.3260	16100	0.0	-
29.4171	16150	0.0	-
29.5082	16200	0.0	-
29.5993	16250	0.0	-
29.6903	16300	0.0	-
29.7814	16350	0.0	-
29.8725	16400	0.0	-
29.9636	16450	0.0	-

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

@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}
}