JohanHeinsen/PE_Kosterbladstagger

SetFit with JohanHeinsen/Old_News_Segmentation_SBERT_V0.1

A model to identify parts of Politiets Efterretninger known as "kosterblade" – i.e. editions describing missing items.

Model Details

Model Description

• Model Type: SetFit
• Sentence Transformer body: JohanHeinsen/Old_News_Segmentation_SBERT_V0.1
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 512 tokens
• Number of Classes: 2 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
1	'Klædningsstykker for Mænd. 33) Benklæder, ny, graastribede, tysk Skrædermrk. Do., brune med svage grønne og lilla Striber, mrk. Torp & Larsen. D. 30 f. M. (St. 3. 966). 34) Vinterfrakke, omtr. ny, graameleret, sildebensvævet, 2 udv. 2 indv. vandr. Lommer, Fløjlskrave, sort Lastingsfoer. D. 29 f. M. (St. 7, 992). 35) Benklæder, lyse, graastribede, Knapperne mrk. Peder Nielsen, København Ø. D. 29 f. M. (St. 7, 975). 36) Jakke, graaternet, toradet, sort Lastingsfoer, 3. udv. og i indv. vandr. Lomme, gul Messingstrop. Cyklesportshue, grønternet, rødt Foer, mrk. "Blaagaardsgade" med forgyldte Bogstaver. D. 28 f.M. (St. 1, 1223). 37) Jakkesæt, blaat Cheviot, enradet, Hornknapper, vandrette Sidelommer, Jakken lang og i FigurSlids bag, mrk. Otto Gøbel, København, brugt. Jakke, blaat Cheviot, toradet, vandrette Sidelommer, 2 indv. Brystlommer, sort Shirtingsfoer. Vest, tilsv., hvid Snor og Kraven, brugt. Siden 27 f. Md. (St. 5, 1216). 38) Jakke, mørkeblaat Cheviot, en Række runde, sorte Hornknapper, vandrette Sidelommer, paa Stroppen et Skrædermrk. Vest, tilsv., hvid Snor om Kraven, brugt. Den 30 f. Md. (St. 5, 1216). 39) Politiuniformsfrakke, ny, mulig uden Knapper eller med ureglementerede Knapper. Siden 5 f. Md. (St. 4). 40) Jakke, mørkeblaat Cheviot, en Række sorte Knapper, sort Foer, 2 lige Sidelommer. Benklæder, tilsv., omtr. nyt. Siden 15 f. Md. (St. 6, 1015). 41) Benklæder, mørkegraa, med noget lysere Striber, en Baglomme, gulligt stribet Linningsfoer. Den 1 ds. (A 7, 1577). 42) Jakke, graaternet, en Rad Hornknapper, ligeLommer, spidse Hjørner, Slids bag, sort Foer, D. 1. ds. (St. 6, 1023). 43) Vinterfrakke, blaa, Fløjlskrave, toradet, brunt, kvadreret Foer. Jakkesæt, brunt med røde og hvide Stænk, enradet. Benklæder, brune. Alt en Del slidt og muli mrk. Brødrene Mæll. D. 30. f.M. (St. 7. 1001). 44) Sommerfrakke, graameleret, Stofkrave, enradet skraa Sidelommer, 2 indv. Brystlommer, brunligt. Lastingsfoer med Striber i Mønstret, Tøjstrop. der er tyndslidt paa Midten, brugt. D. 1. ds. (St. 5, 1229). 45) Jakkesæt, mørkeblaat Cheviot, enradet, alm. Lommer, sorte Hornknapper, brugt. D. 1. ds. (St. 5, 1221). 46) Jakke, mørkeblaat Cheviot, 2 Rækker Vachtklubknapper, alm. Lommer, brugt. Den 1. ds. (St. 5, 1222).' 'Linned Sengklæder og desl. 24) Lagener, et Par Dovlas til en enkelt Seng, mrk. B. W., syet med hvidt Garn. Den 5/5. (St. 1, 540). 25) Skjorte, aldeles ny Tvistlærreds. Den 8/5. (St. 2, 535).' 'Linned, Sengklæder og dsl. 14) Lagen, af Dovlas, til en enkelt Seng, mrk. med sort Syning L. D. D. 10/1. (St. 2.'
0	'Afleverede til Fattigvæsenet. Frederik Ferdinand Sørensen (Kjøbenh.), 45 Aar. Emma Emillie Hansen (Kjøbenh.), 19 Aar. Johanne Christiane Scheele, ofl. Fr. 22. Aar. Ane Sophie Larsen Allerup (Kjøbenh.), 41 Aar. Christine Petersen (Uggeløse), 17 Aar.' 'Løsladte efter udstaaet Straf. Andreas Jensen (Nibe), 58 Aar. 8 D. simp. F. for Betleri. Delfingade 23. [6797]. Martin Vilhelm Jensen (Randers), 38 Aar. 8 D. simp. F. for Betleri. Thuresensgade 4, St. [6798]. Carl Ferdinand Gottliebsen (Kbhvn.), 40 Aar. D. simp. F. for Betleri. Landemærket 51. 16799]. Sophie Albertine Albrechtsen (Kbhvn.), 37 Aar. 8 x 5 D. F. p. V. o. B. for Bedrageri. [2152]. Carl Johan Sørensen (Kbhvn.), 23 Aar. 30 D. Tvangsarb. for Løsgængeri. Skyttegade 11. [445]. Poul Christian Petersen (Rorup), 69 Aar. 24 D. Tvangsarb. for Betleri. Afl. til Fattv. [2638]. Mads Kjeldsen (Ledøje), 45 Aar. 24 D. Tvangsarb. for Betleri. Afl. til Fattv. [701]. Bodil Kirstine Dorthea Thomsen (Tranegilde), 21 Aar. 3 x 5 D. F. p. V. o. B. for Overtr. af Strls. § 181. Dimitt. [1403]. Louis Ferdinand Thorvald Jensen (Kbhvn.), 20 Aar. 24 D. Tvangsarb. for Betleri. Arrest. under en ny Sag. [5745].' 'Af Retten Løsladte. Ane Cathrine Olesdatter (Axelholm), 56 Aar. Nansensgade 53, 1. Sal. Victor Vilhelm Nylund (Abo, Finland), 29 Aar. Carl Christian Larsen (Villingebæk), 43 Aar. Christianshavns Torv 6, Bgst. [539]. Andreas Peter Christian Deckmann (Rendsburg), 44 Aar. St. Hansgade 10, 4. Sal. [365]. Hans Jørgensen (Norge), 24 Aar. Dimitt. [538]. Gunder Johansen (Norge), 27 Aar. Dimitt. [537].'

Evaluation

Metrics

Label	Accuracy	F1	Precision	Recall
all	0.9992	0.9990	1.0	0.9979

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("setfit_model_id")
# Run inference
preds = model("Anholdte for Forbrydelser m.V. Stanislaus Wladislaus v. Bogutezinsk, (Warschau, 23 Aar. Bedrageri, August Nikolai Thomsen (Kbhvn.), 20 Aar. Grovt Tyveri. Ingvard Frederik Møller (Kbhvn.), 49 Aar. Tyveri. og Løsgængeri. Jens Peter Nielsen (Sørninge), 26 Aar. Sigtet for grovt Tyveri. Heinrich Jäger (Kiel), 23 Aar. Løsgængeri.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	12	150.3843	1357

Label	Training Sample Count
0	8397
1	5603

Training Hyperparameters

• batch_size: (12, 12)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 8
• 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: 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
0.0001	1	0.2330	-
0.0027	50	0.1617	-
0.0054	100	0.1177	-
0.0080	150	0.0607	-
0.0107	200	0.0316	-
0.0134	250	0.0137	-
0.0161	300	0.0138	-
0.0187	350	0.0078	-
0.0214	400	0.0019	-
0.0241	450	0.0088	-
0.0268	500	0.0041	-
0.0295	550	0.0016	-
0.0321	600	0.0077	-
0.0348	650	0.0044	-
0.0375	700	0.0009	-
0.0402	750	0.0028	-
0.0429	800	0.0013	-
0.0455	850	0.0008	-
0.0482	900	0.0023	-
0.0509	950	0.0034	-
0.0536	1000	0.0042	-
0.0562	1050	0.0013	-
0.0589	1100	0.0030	-
0.0616	1150	0.0017	-
0.0643	1200	0.0023	-
0.0670	1250	0.0039	-
0.0696	1300	0.0015	-
0.0723	1350	0.0001	-
0.0750	1400	0.0029	-
0.0777	1450	0.0021	-
0.0804	1500	0.0022	-
0.0830	1550	0.0035	-
0.0857	1600	0.0016	-
0.0884	1650	0.0011	-
0.0911	1700	0.0062	-
0.0937	1750	0.0020	-
0.0964	1800	0.0029	-
0.0991	1850	0.0015	-
0.1018	1900	0.0002	-
0.1045	1950	0.0016	-
0.1071	2000	0.0000	-
0.1098	2050	0.0000	-
0.1125	2100	0.0045	-
0.1152	2150	0.0001	-
0.1179	2200	0.0003	-
0.1205	2250	0.0000	-
0.1232	2300	0.0000	-
0.1259	2350	0.0000	-
0.1286	2400	0.0000	-
0.1312	2450	0.0000	-
0.1339	2500	0.0000	-
0.1366	2550	0.0083	-
0.1393	2600	0.0020	-
0.1420	2650	0.0001	-
0.1446	2700	0.0022	-
0.1473	2750	0.0035	-
0.1500	2800	0.0003	-
0.1527	2850	0.0001	-
0.1554	2900	0.0000	-
0.1580	2950	0.0000	-
0.1607	3000	0.0043	-
0.1634	3050	0.0000	-
0.1661	3100	0.0000	-
0.1687	3150	0.0000	-
0.1714	3200	0.0017	-
0.1741	3250	0.0002	-
0.1768	3300	0.0000	-
0.1795	3350	0.0000	-
0.1821	3400	0.0000	-
0.1848	3450	0.0000	-
0.1875	3500	0.0000	-
0.1902	3550	0.0000	-
0.1929	3600	0.0000	-
0.1955	3650	0.0001	-
0.1982	3700	0.0000	-
0.2009	3750	0.0000	-
0.2036	3800	0.0000	-
0.2062	3850	0.0000	-
0.2089	3900	0.0000	-
0.2116	3950	0.0000	-
0.2143	4000	0.0000	-
0.2170	4050	0.0010	-
0.2196	4100	0.0003	-
0.2223	4150	0.0017	-
0.2250	4200	0.0000	-
0.2277	4250	0.0050	-
0.2304	4300	0.0050	-
0.2330	4350	0.0005	-
0.2357	4400	0.0000	-
0.2384	4450	0.0008	-
0.2411	4500	0.0000	-
0.2437	4550	0.0000	-
0.2464	4600	0.0019	-
0.2491	4650	0.0021	-
0.2518	4700	0.0002	-
0.2545	4750	0.0000	-
0.2571	4800	0.0000	-
0.2598	4850	0.0000	-
0.2625	4900	0.0000	-
0.2652	4950	0.0001	-
0.2679	5000	0.0000	-
0.2705	5050	0.0000	-
0.2732	5100	0.0000	-
0.2759	5150	0.0000	-
0.2786	5200	0.0000	-
0.2812	5250	0.0000	-
0.2839	5300	0.0000	-
0.2866	5350	0.0000	-
0.2893	5400	0.0000	-
0.2920	5450	0.0000	-
0.2946	5500	0.0000	-
0.2973	5550	0.0000	-
0.3000	5600	0.0000	-
0.3027	5650	0.0000	-
0.3054	5700	0.0000	-
0.3080	5750	0.0000	-
0.3107	5800	0.0000	-
0.3134	5850	0.0000	-
0.3161	5900	0.0000	-
0.3187	5950	0.0000	-
0.3214	6000	0.0000	-
0.3241	6050	0.0000	-
0.3268	6100	0.0000	-
0.3295	6150	0.0000	-
0.3321	6200	0.0000	-
0.3348	6250	0.0000	-
0.3375	6300	0.0000	-
0.3402	6350	0.0000	-
0.3429	6400	0.0000	-
0.3455	6450	0.0000	-
0.3482	6500	0.0000	-
0.3509	6550	0.0000	-
0.3536	6600	0.0000	-
0.3562	6650	0.0000	-
0.3589	6700	0.0000	-
0.3616	6750	0.0000	-
0.3643	6800	0.0000	-
0.3670	6850	0.0000	-
0.3696	6900	0.0000	-
0.3723	6950	0.0000	-
0.3750	7000	0.0000	-
0.3777	7050	0.0000	-
0.3804	7100	0.0000	-
0.3830	7150	0.0000	-
0.3857	7200	0.0000	-
0.3884	7250	0.0000	-
0.3911	7300	0.0000	-
0.3937	7350	0.0000	-
0.3964	7400	0.0000	-
0.3991	7450	0.0000	-
0.4018	7500	0.0000	-
0.4045	7550	0.0000	-
0.4071	7600	0.0000	-
0.4098	7650	0.0000	-
0.4125	7700	0.0000	-
0.4152	7750	0.0000	-
0.4178	7800	0.0000	-
0.4205	7850	0.0000	-
0.4232	7900	0.0000	-
0.4259	7950	0.0000	-
0.4286	8000	0.0000	-
0.4312	8050	0.0000	-
0.4339	8100	0.0000	-
0.4366	8150	0.0000	-
0.4393	8200	0.0000	-
0.4420	8250	0.0000	-
0.4446	8300	0.0000	-
0.4473	8350	0.0000	-
0.4500	8400	0.0000	-
0.4527	8450	0.0000	-
0.4553	8500	0.0000	-
0.4580	8550	0.0000	-
0.4607	8600	0.0000	-
0.4634	8650	0.0000	-
0.4661	8700	0.0000	-
0.4687	8750	0.0000	-
0.4714	8800	0.0000	-
0.4741	8850	0.0000	-
0.4768	8900	0.0000	-
0.4795	8950	0.0000	-
0.4821	9000	0.0000	-
0.4848	9050	0.0000	-
0.4875	9100	0.0000	-
0.4902	9150	0.0000	-
0.4928	9200	0.0000	-
0.4955	9250	0.0000	-
0.4982	9300	0.0000	-
0.5009	9350	0.0000	-
0.5036	9400	0.0000	-
0.5062	9450	0.0000	-
0.5089	9500	0.0000	-
0.5116	9550	0.0000	-
0.5143	9600	0.0000	-
0.5170	9650	0.0000	-
0.5196	9700	0.0000	-
0.5223	9750	0.0000	-
0.5250	9800	0.0000	-
0.5277	9850	0.0000	-
0.5303	9900	0.0000	-
0.5330	9950	0.0000	-
0.5357	10000	0.0000	-
0.5384	10050	0.0000	-
0.5411	10100	0.0000	-
0.5437	10150	0.0000	-
0.5464	10200	0.0000	-
0.5491	10250	0.0000	-
0.5518	10300	0.0000	-
0.5545	10350	0.0000	-
0.5571	10400	0.0000	-
0.5598	10450	0.0000	-
0.5625	10500	0.0000	-
0.5652	10550	0.0000	-
0.5678	10600	0.0000	-
0.5705	10650	0.0000	-
0.5732	10700	0.0000	-
0.5759	10750	0.0000	-
0.5786	10800	0.0000	-
0.5812	10850	0.0000	-
0.5839	10900	0.0000	-
0.5866	10950	0.0000	-
0.5893	11000	0.0000	-
0.5920	11050	0.0000	-
0.5946	11100	0.0000	-
0.5973	11150	0.0000	-
0.6000	11200	0.0000	-
0.6027	11250	0.0000	-
0.6053	11300	0.0000	-
0.6080	11350	0.0000	-
0.6107	11400	0.0000	-
0.6134	11450	0.0000	-
0.6161	11500	0.0000	-
0.6187	11550	0.0000	-
0.6214	11600	0.0000	-
0.6241	11650	0.0000	-
0.6268	11700	0.0000	-
0.6295	11750	0.0000	-
0.6321	11800	0.0000	-
0.6348	11850	0.0000	-
0.6375	11900	0.0000	-
0.6402	11950	0.0000	-
0.6428	12000	0.0000	-
0.6455	12050	0.0000	-
0.6482	12100	0.0000	-
0.6509	12150	0.0000	-
0.6536	12200	0.0000	-
0.6562	12250	0.0000	-
0.6589	12300	0.0000	-
0.6616	12350	0.0000	-
0.6643	12400	0.0000	-
0.6670	12450	0.0000	-
0.6696	12500	0.0000	-
0.6723	12550	0.0000	-
0.6750	12600	0.0000	-
0.6777	12650	0.0000	-
0.6803	12700	0.0000	-
0.6830	12750	0.0000	-
0.6857	12800	0.0000	-
0.6884	12850	0.0000	-
0.6911	12900	0.0000	-
0.6937	12950	0.0000	-
0.6964	13000	0.0000	-
0.6991	13050	0.0000	-
0.7018	13100	0.0000	-
0.7045	13150	0.0000	-
0.7071	13200	0.0000	-
0.7098	13250	0.0000	-
0.7125	13300	0.0000	-
0.7152	13350	0.0000	-
0.7178	13400	0.0000	-
0.7205	13450	0.0000	-
0.7232	13500	0.0000	-
0.7259	13550	0.0000	-
0.7286	13600	0.0000	-
0.7312	13650	0.0000	-
0.7339	13700	0.0000	-
0.7366	13750	0.0000	-
0.7393	13800	0.0000	-
0.7420	13850	0.0000	-
0.7446	13900	0.0000	-
0.7473	13950	0.0000	-
0.7500	14000	0.0000	-
0.7527	14050	0.0000	-
0.7553	14100	0.0000	-
0.7580	14150	0.0000	-
0.7607	14200	0.0000	-
0.7634	14250	0.0000	-
0.7661	14300	0.0000	-
0.7687	14350	0.0000	-
0.7714	14400	0.0000	-
0.7741	14450	0.0000	-
0.7768	14500	0.0000	-
0.7795	14550	0.0000	-
0.7821	14600	0.0000	-
0.7848	14650	0.0000	-
0.7875	14700	0.0000	-
0.7902	14750	0.0000	-
0.7928	14800	0.0000	-
0.7955	14850	0.0000	-
0.7982	14900	0.0000	-
0.8009	14950	0.0000	-
0.8036	15000	0.0000	-
0.8062	15050	0.0000	-
0.8089	15100	0.0000	-
0.8116	15150	0.0000	-
0.8143	15200	0.0000	-
0.8169	15250	0.0000	-
0.8196	15300	0.0000	-
0.8223	15350	0.0000	-
0.8250	15400	0.0000	-
0.8277	15450	0.0000	-
0.8303	15500	0.0000	-
0.8330	15550	0.0000	-
0.8357	15600	0.0000	-
0.8384	15650	0.0000	-
0.8411	15700	0.0000	-
0.8437	15750	0.0000	-
0.8464	15800	0.0000	-
0.8491	15850	0.0000	-
0.8518	15900	0.0000	-
0.8544	15950	0.0000	-
0.8571	16000	0.0000	-
0.8598	16050	0.0000	-
0.8625	16100	0.0000	-
0.8652	16150	0.0000	-
0.8678	16200	0.0000	-
0.8705	16250	0.0000	-
0.8732	16300	0.0000	-
0.8759	16350	0.0000	-
0.8786	16400	0.0000	-
0.8812	16450	0.0000	-
0.8839	16500	0.0000	-
0.8866	16550	0.0000	-
0.8893	16600	0.0000	-
0.8919	16650	0.0000	-
0.8946	16700	0.0000	-
0.8973	16750	0.0000	-
0.9000	16800	0.0000	-
0.9027	16850	0.0000	-
0.9053	16900	0.0000	-
0.9080	16950	0.0000	-
0.9107	17000	0.0000	-
0.9134	17050	0.0000	-
0.9161	17100	0.0000	-
0.9187	17150	0.0000	-
0.9214	17200	0.0000	-
0.9241	17250	0.0000	-
0.9268	17300	0.0000	-
0.9294	17350	0.0000	-
0.9321	17400	0.0000	-
0.9348	17450	0.0000	-
0.9375	17500	0.0000	-
0.9402	17550	0.0000	-
0.9428	17600	0.0000	-
0.9455	17650	0.0000	-
0.9482	17700	0.0000	-
0.9509	17750	0.0000	-
0.9536	17800	0.0000	-
0.9562	17850	0.0000	-
0.9589	17900	0.0000	-
0.9616	17950	0.0000	-
0.9643	18000	0.0000	-
0.9669	18050	0.0000	-
0.9696	18100	0.0000	-
0.9723	18150	0.0000	-
0.9750	18200	0.0000	-
0.9777	18250	0.0000	-
0.9803	18300	0.0000	-
0.9830	18350	0.0000	-
0.9857	18400	0.0000	-
0.9884	18450	0.0000	-
0.9911	18500	0.0000	-
0.9937	18550	0.0000	-
0.9964	18600	0.0000	-
0.9991	18650	0.0000	-

Framework Versions

• Python: 3.11.12
• SetFit: 1.2.0
• Sentence Transformers: 6.2.0
• Transformers: 5.20.2
• PyTorch: 2.10.0
• Datasets: 5.1.0
• Tokenizers: 0.23.3

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}
}