This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-MiniLM-L6-v2 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: sentence-transformers/paraphrase-MiniLM-L6-v2
Classification head: a LogisticRegression instance
Maximum Sequence Length: 128 tokens
Number of Classes: 9 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
navigation	'https://www.groupe-ocea.fr' 'https://woemen.fr/' 'https://fourchenoire.com/fr'
product	'https://www.woemen.fr/products/tshirt-oversize-mixte-blanc' 'https://www.bivouack.fr/salon-sejour/lit-escamotable-canape/ensemble-lit-escamotable-horizontal-avec-canape-et-bureau-citala' 'https://www.bivouack.fr/lit-escamotable-vertical-nida'
service	'https://pub-pro.fr/service/decorations-adhesives-depolies/' 'https://bclimatisation.fr/installation-chauffage-rennes/' 'https://www.philartstudio.com/mariages/reportage/noelie-marc/'
brand	'https://ynspir.com/guide-marques/orac-decor/' 'https://cogitech.fr/en/studio-2/' 'https://woemen.fr/pages/woemen-x-lubia-partenariat-protections-periodiques'
legal	'https://ditesmoioui.com/faqs/' 'https://www.scal-ia.fr/contact' 'https://pub-pro.fr/contact'
listing	'https://www.pepejeans.com/fr_fr/femme/accessoires/lunettes-de-soleil' 'https://ynspir.com/conseils-decoration/agencement/amenagement-studio/' 'https://www.ebspaca.fr/service-carousel/'
post	'https://woemen.fr/blogs/le-coton-biologique' 'https://ynspir.com/divers/quelles-couleurs-choisir-pour-un-interieur-lumineux-et-chaleureux/' 'https://www.lacompagnieducaviar.fr/post/nos-idées-de-recettes-comment-sublimer-vos-plats-avec-nos-produits-autour-du-caviar'
testimonial	'https://cogitech.fr/en/realisation/design-en/epee/' 'https://cogitech.fr/realisation/design/dynamic-landscape-fr/' 'https://cogitech.fr/en/realisation/design-en/apollo-2/'
account_shop	'https://www.pepejeans.com/fr_fr/jeans-fit-guide-man.html' 'https://www.woemen.fr/policies/refund-policy' 'https://lesjupesdeprune.com/customer_authentication/login'

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("Wispra-fr/setfit-url-classifier")
# Run inference
preds = model("https://ynspir.com")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	1.0	1

Label	Training Sample Count
legal	70
brand	86
product	176
service	184
post	133
navigation	53
listing	265
account_shop	38
testimonial	132

Training Hyperparameters

batch_size: (32, 32)
num_epochs: (5, 5)
max_steps: -1
sampling_strategy: undersampling
num_iterations: 10
body_learning_rate: (2e-05, 2e-05)
head_learning_rate: 2e-05
loss: CosineSimilarityLoss
distance_metric: cosine_distance
margin: 0.25
end_to_end: False
use_amp: True
warmup_proportion: 0.1
max_length: 128
seed: 42
eval_max_steps: -1
load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0014	1	0.2637	-
0.0141	10	0.2127	-
0.0281	20	0.2137	-
0.0422	30	0.2431	-
0.0563	40	0.2347	-
0.0703	50	0.2264	-
0.0844	60	0.2391	-
0.0985	70	0.2395	-
0.1125	80	0.253	-
0.1266	90	0.2295	-
0.1406	100	0.2281	-
0.1547	110	0.2216	-
0.1688	120	0.2297	-
0.1828	130	0.2471	-
0.1969	140	0.209	-
0.2110	150	0.1998	-
0.2250	160	0.1932	-
0.2391	170	0.1821	-
0.2532	180	0.1894	-
0.2672	190	0.1685	-
0.2813	200	0.163	-
0.2954	210	0.2206	-
0.3094	220	0.1698	-
0.3235	230	0.2171	-
0.3376	240	0.1834	-
0.3516	250	0.1678	-
0.3657	260	0.1537	-
0.3797	270	0.1527	-
0.3938	280	0.1751	-
0.4079	290	0.152	-
0.4219	300	0.1371	-
0.4360	310	0.1403	-
0.4501	320	0.1042	-
0.4641	330	0.1108	-
0.4782	340	0.1003	-
0.4923	350	0.1226	-
0.5063	360	0.1613	-
0.5204	370	0.1259	-
0.5345	380	0.0877	-
0.5485	390	0.1067	-
0.5626	400	0.1143	-
0.5767	410	0.096	-
0.5907	420	0.0557	-
0.6048	430	0.051	-
0.6188	440	0.1339	-
0.6329	450	0.0846	-
0.6470	460	0.0657	-
0.6610	470	0.0812	-
0.6751	480	0.058	-
0.6892	490	0.093	-
0.7032	500	0.0397	-
0.7173	510	0.0932	-
0.7314	520	0.062	-
0.7454	530	0.0595	-
0.7595	540	0.0774	-
0.7736	550	0.0444	-
0.7876	560	0.06	-
0.8017	570	0.0486	-
0.8158	580	0.0708	-
0.8298	590	0.0518	-
0.8439	600	0.0479	-
0.8579	610	0.0511	-
0.8720	620	0.0722	-
0.8861	630	0.0691	-
0.9001	640	0.0841	-
0.9142	650	0.0997	-
0.9283	660	0.0583	-
0.9423	670	0.0264	-
0.9564	680	0.0452	-
0.9705	690	0.0174	-
0.9845	700	0.0448	-
0.9986	710	0.043	-
1.0127	720	0.0801	-
1.0267	730	0.063	-
1.0408	740	0.0376	-
1.0549	750	0.0273	-
1.0689	760	0.0516	-
1.0830	770	0.0317	-
1.0970	780	0.0247	-
1.1111	790	0.0254	-
1.1252	800	0.0213	-
1.1392	810	0.0172	-
1.1533	820	0.0365	-
1.1674	830	0.0247	-
1.1814	840	0.0471	-
1.1955	850	0.0248	-
1.2096	860	0.0711	-
1.2236	870	0.0231	-
1.2377	880	0.0504	-
1.2518	890	0.0477	-
1.2658	900	0.0104	-
1.2799	910	0.0338	-
1.2940	920	0.0116	-
1.3080	930	0.0599	-
1.3221	940	0.0215	-
1.3361	950	0.0418	-
1.3502	960	0.0265	-
1.3643	970	0.0371	-
1.3783	980	0.0334	-
1.3924	990	0.0478	-
1.4065	1000	0.0223	-
1.4205	1010	0.0132	-
1.4346	1020	0.0166	-
1.4487	1030	0.035	-
1.4627	1040	0.0081	-
1.4768	1050	0.0238	-
1.4909	1060	0.0177	-
1.5049	1070	0.009	-
1.5190	1080	0.0296	-
1.5331	1090	0.0323	-
1.5471	1100	0.0237	-
1.5612	1110	0.0298	-
1.5752	1120	0.0592	-
1.5893	1130	0.0052	-
1.6034	1140	0.0112	-
1.6174	1150	0.0477	-
1.6315	1160	0.0356	-
1.6456	1170	0.0324	-
1.6596	1180	0.0412	-
1.6737	1190	0.0484	-
1.6878	1200	0.0262	-
1.7018	1210	0.011	-
1.7159	1220	0.0075	-
1.7300	1230	0.0471	-
1.7440	1240	0.0398	-
1.7581	1250	0.0335	-
1.7722	1260	0.0278	-
1.7862	1270	0.0533	-
1.8003	1280	0.0291	-
1.8143	1290	0.0122	-
1.8284	1300	0.0039	-
1.8425	1310	0.0043	-
1.8565	1320	0.0135	-
1.8706	1330	0.0182	-
1.8847	1340	0.0306	-
1.8987	1350	0.0135	-
1.9128	1360	0.0034	-
1.9269	1370	0.0109	-
1.9409	1380	0.0209	-
1.9550	1390	0.0244	-
1.9691	1400	0.0052	-
1.9831	1410	0.0095	-
1.9972	1420	0.0067	-
2.0113	1430	0.0091	-
2.0253	1440	0.0077	-
2.0394	1450	0.0246	-
2.0534	1460	0.0123	-
2.0675	1470	0.0061	-
2.0816	1480	0.0375	-
2.0956	1490	0.0187	-
2.1097	1500	0.0029	-
2.1238	1510	0.0043	-
2.1378	1520	0.0191	-
2.1519	1530	0.0039	-
2.1660	1540	0.0628	-
2.1800	1550	0.0278	-
2.1941	1560	0.0106	-
2.2082	1570	0.0192	-
2.2222	1580	0.0127	-
2.2363	1590	0.0053	-
2.2504	1600	0.0211	-
2.2644	1610	0.0291	-
2.2785	1620	0.0043	-
2.2925	1630	0.0147	-
2.3066	1640	0.0219	-
2.3207	1650	0.0017	-
2.3347	1660	0.0114	-
2.3488	1670	0.0056	-
2.3629	1680	0.0075	-
2.3769	1690	0.0191	-
2.3910	1700	0.0049	-
2.4051	1710	0.0279	-
2.4191	1720	0.0081	-
2.4332	1730	0.0047	-
2.4473	1740	0.0035	-
2.4613	1750	0.0024	-
2.4754	1760	0.0022	-
2.4895	1770	0.0091	-
2.5035	1780	0.0238	-
2.5176	1790	0.0084	-
2.5316	1800	0.0267	-
2.5457	1810	0.0071	-
2.5598	1820	0.0027	-
2.5738	1830	0.0226	-
2.5879	1840	0.0032	-
2.6020	1850	0.0014	-
2.6160	1860	0.0028	-
2.6301	1870	0.0043	-
2.6442	1880	0.0105	-
2.6582	1890	0.0036	-
2.6723	1900	0.0031	-
2.6864	1910	0.008	-
2.7004	1920	0.0296	-
2.7145	1930	0.0103	-
2.7286	1940	0.0234	-
2.7426	1950	0.0035	-
2.7567	1960	0.0252	-
2.7707	1970	0.0238	-
2.7848	1980	0.0045	-
2.7989	1990	0.0304	-
2.8129	2000	0.0021	-
2.8270	2010	0.0046	-
2.8411	2020	0.0027	-
2.8551	2030	0.0169	-
2.8692	2040	0.0089	-
2.8833	2050	0.0187	-
2.8973	2060	0.0032	-
2.9114	2070	0.0025	-
2.9255	2080	0.0161	-
2.9395	2090	0.0023	-
2.9536	2100	0.0014	-
2.9677	2110	0.004	-
2.9817	2120	0.0061	-
2.9958	2130	0.0227	-
3.0098	2140	0.0012	-
3.0239	2150	0.0377	-
3.0380	2160	0.0145	-
3.0520	2170	0.022	-
3.0661	2180	0.0017	-
3.0802	2190	0.0013	-
3.0942	2200	0.0018	-
3.1083	2210	0.0025	-
3.1224	2220	0.0024	-
3.1364	2230	0.0088	-
3.1505	2240	0.0019	-
3.1646	2250	0.0159	-
3.1786	2260	0.004	-
3.1927	2270	0.0008	-
3.2068	2280	0.0031	-
3.2208	2290	0.0037	-
3.2349	2300	0.0015	-
3.2489	2310	0.0041	-
3.2630	2320	0.0026	-
3.2771	2330	0.0009	-
3.2911	2340	0.0022	-
3.3052	2350	0.0028	-
3.3193	2360	0.0161	-
3.3333	2370	0.0076	-
3.3474	2380	0.0021	-
3.3615	2390	0.0172	-
3.3755	2400	0.0292	-
3.3896	2410	0.0051	-
3.4037	2420	0.0174	-
3.4177	2430	0.0081	-
3.4318	2440	0.0065	-
3.4459	2450	0.0024	-
3.4599	2460	0.0094	-
3.4740	2470	0.0146	-
3.4880	2480	0.004	-
3.5021	2490	0.0044	-
3.5162	2500	0.0016	-
3.5302	2510	0.0032	-
3.5443	2520	0.0289	-
3.5584	2530	0.0123	-
3.5724	2540	0.0055	-
3.5865	2550	0.0031	-
3.6006	2560	0.0081	-
3.6146	2570	0.0042	-
3.6287	2580	0.0051	-
3.6428	2590	0.0058	-
3.6568	2600	0.0017	-
3.6709	2610	0.005	-
3.6850	2620	0.0015	-
3.6990	2630	0.0008	-
3.7131	2640	0.0069	-
3.7271	2650	0.0022	-
3.7412	2660	0.0024	-
3.7553	2670	0.0018	-
3.7693	2680	0.0031	-
3.7834	2690	0.0112	-
3.7975	2700	0.0051	-
3.8115	2710	0.0024	-
3.8256	2720	0.0011	-
3.8397	2730	0.0008	-
3.8537	2740	0.0035	-
3.8678	2750	0.0029	-
3.8819	2760	0.0047	-
3.8959	2770	0.0208	-
3.9100	2780	0.0026	-
3.9241	2790	0.0152	-
3.9381	2800	0.0021	-
3.9522	2810	0.0188	-
3.9662	2820	0.0162	-
3.9803	2830	0.0009	-
3.9944	2840	0.0045	-
4.0084	2850	0.0058	-
4.0225	2860	0.031	-
4.0366	2870	0.0013	-
4.0506	2880	0.0021	-
4.0647	2890	0.0022	-
4.0788	2900	0.008	-
4.0928	2910	0.0107	-
4.1069	2920	0.0015	-
4.1210	2930	0.003	-
4.1350	2940	0.0094	-
4.1491	2950	0.0013	-
4.1632	2960	0.0084	-
4.1772	2970	0.0021	-
4.1913	2980	0.0068	-
4.2053	2990	0.0032	-
4.2194	3000	0.0044	-
4.2335	3010	0.0037	-
4.2475	3020	0.0182	-
4.2616	3030	0.0023	-
4.2757	3040	0.0019	-
4.2897	3050	0.001	-
4.3038	3060	0.0149	-
4.3179	3070	0.0241	-
4.3319	3080	0.0028	-
4.3460	3090	0.0104	-
4.3601	3100	0.0007	-
4.3741	3110	0.0015	-
4.3882	3120	0.0023	-
4.4023	3130	0.0228	-
4.4163	3140	0.0161	-
4.4304	3150	0.0151	-
4.4444	3160	0.0043	-
4.4585	3170	0.0031	-
4.4726	3180	0.0041	-
4.4866	3190	0.0006	-
4.5007	3200	0.005	-
4.5148	3210	0.0027	-
4.5288	3220	0.0019	-
4.5429	3230	0.003	-
4.5570	3240	0.0024	-
4.5710	3250	0.0167	-
4.5851	3260	0.001	-
4.5992	3270	0.0022	-
4.6132	3280	0.013	-
4.6273	3290	0.0095	-
4.6414	3300	0.0202	-
4.6554	3310	0.0147	-
4.6695	3320	0.0009	-
4.6835	3330	0.0008	-
4.6976	3340	0.009	-
4.7117	3350	0.0018	-
4.7257	3360	0.0043	-
4.7398	3370	0.0014	-
4.7539	3380	0.0015	-
4.7679	3390	0.0111	-
4.7820	3400	0.0028	-
4.7961	3410	0.0019	-
4.8101	3420	0.0005	-
4.8242	3430	0.0102	-
4.8383	3440	0.0015	-
4.8523	3450	0.0014	-
4.8664	3460	0.0007	-
4.8805	3470	0.0006	-
4.8945	3480	0.0022	-
4.9086	3490	0.0034	-
4.9226	3500	0.0005	-
4.9367	3510	0.0055	-
4.9508	3520	0.0013	-
4.9648	3530	0.003	-
4.9789	3540	0.0105	-
4.9930	3550	0.0007	-

Framework Versions

Python: 3.11.0rc1
SetFit: 1.0.3
Sentence Transformers: 2.3.1
Transformers: 4.40.2
PyTorch: 2.1.2+cu121
Datasets: 2.16.1
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}
}