Характеристики

Лицензия: CC BY 4.0
Базовая модель: xlm-roberta-large
Библиотека: transformers
Опубликована: 9 мар 2026

Ссылки

Hugging Face

xlm-roberta-large-pooled-cap-minor-v5

poltextlab

Классификация текстаТекст → label3 профессии

Дообученная многоязычная XLM-RoBERTa Large классифицирует тексты по темам CAP Minor для сравнительных исследований.

56скачиваний за 30 дней1лайквеса обновлены24 авг 2026

Бенчмарки

Как модель справилась с задачами пользователей ailist и что о ней публикуют открытые лидерборды

Публичные оценки

Из открытых лидербордов. Сравнивайте результаты одного теста при одинаковых условиях — отсутствие оценки не значит, что модель хуже.

В подключённых источниках результатов для этой модели пока нет.

Что умеет модель

Описание составлено по карточке модели и метаданным Hugging Face

Модель poltextlab/xlm-roberta-large-pooled-cap-minor-v5 разработана PoltextLab и представляет собой дообученную версию XLM-RoBERTa Large для многоклассовой классификации текстов по темам CAP Minor. Она предназначена для разметки многоязычных текстов в сравнительных исследованиях и используется в инструменте CAP Babel Machine.

Модель поддерживает английский, венгерский, португальский, датский, французский, итальянский и испанский языки. В карточке приведены метрики по языкам: взвешенный F1 составляет 0,88 для венгерского и 0,86 для португальского, тогда как для датского, французского и испанского он ниже. Для всех классов макроусреднённый F1 равен 0,63, взвешенный F1 — 0,81.

Модель распространяется в формате Safetensors и доступна по лицензии CC BY 4.0. Результаты зависят от языка и темы; приведённые метрики показывают заметно более низкое качество для части поддерживаемых языков.

Сценарии применения

Разметка текстов по темам CAP Minor
Классификация новостных материалов по макроэкономическим темам
Подготовка многоязычных корпусов для сравнительных исследований
Автоматическая тематическая разметка текстов на английском и европейских языках

Ограничения

Качество заметно различается по языкам: взвешенный F1 составляет 0,48 для датского, 0,50 для французского и 0,52 для испанского. Общий макроусреднённый F1 равен 0,63.

Для каких профессий подходит3

Разделы каталога профессий, в задачах которых используется эта модель

Все профессии

I.11Универсальные способностиКлассификация, ранжирование и сопоставлениеAI помогает распределять объекты по категориям, устанавливать соответствия и определять порядок обработки.15 задач26 736моделей I.3Универсальные способностиИзвлечение и структурирование данныхAI помогает превращать разрозненные данные в структуру, пригодную для поиска, аналитики и автоматизации.18 задач25 791модель IV.9Отраслевые задачиМедиа, журналистика, культура, архивы и игрыAI помогает создавать, искать, сохранять и персонализировать медиа, культурный и игровой контент.44 задачи14 938моделей

Карточка модели

README от автора на Hugging Face, как он его опубликовал

Открыть на Hugging Face

xlm-roberta-large-pooled-cap-minor-v5

This model was fine-tuned on multilingual multi-domain data labeled with the CAP Minor topics (https://www.comparativeagendas.net/pages/master-codebook).

How to use the model

from transformers import AutoTokenizer, pipeline

tokenizer = AutoTokenizer.from_pretrained("xlm-roberta-large")
pipe = pipeline(
    model="poltextlab/xlm-roberta-large-pooled-cap-minor-v5-e8checkpoint",
    task="text-classification",
    tokenizer=tokenizer,
    use_fast=False,
    token="<your_hf_read_only_token>"
)

text = "<text_to_classify>"
pipe(text)

Classification Report

Overall Performance:

Accuracy: N/A
Macro Avg: Precision: 0.68, Recall: 0.61, F1-score: 0.63
Weighted Avg: Precision: 0.81, Recall: 0.82, F1-score: 0.81

Per-Language Weighted F1-Scores:

hu: 0.88 (on 204 labels)
pt: 0.86 (on 146 labels)
en: 0.79 (on 213 labels)
it: 0.56 (on 193 labels)
es: 0.52 (on 187 labels)
fr: 0.50 (on 192 labels)
da: 0.48 (on 130 labels)

Per-Class Metrics:

Label	Precision	Recall	F1-score	Support
(100) Macroeconomics – General	0.64	0.67	0.65	2378
(101) Macroeconomics – Interest Rates	0.71	0.61	0.66	313
(103) Macroeconomics – Unemployment Rate	0.59	0.52	0.55	454
(104) Macroeconomics – Monetary Policy	0.72	0.68	0.7	645
(105) Macroeconomics – National Budget	0.75	0.81	0.78	5394
(107) Macroeconomics – Tax Code	0.74	0.79	0.77	3152
(108) Macroeconomics – Industrial Policy	0.65	0.65	0.65	871
(110) Macroeconomics – Price Control	0.62	0.48	0.54	160
(199) Macroeconomics – Other	0	0	0	15
(200) Civil Rights – General	0.59	0.59	0.59	796
(201) Civil Rights – Minority Discrimination	0.74	0.72	0.73	723
(202) Civil Rights – Gender Discrimination	0.75	0.75	0.75	554
(204) Civil Rights – Age Discrimination	0.64	0.33	0.44	54
(205) Civil Rights – Handicap Discrimination	0.74	0.66	0.69	152
(206) Civil Rights – Voting Rights	0.71	0.65	0.68	284
(207) Civil Rights – Freedom of Speech	0.72	0.67	0.69	389
(208) Civil Rights – Right to Privacy	0.71	0.62	0.66	561
(209) Civil Rights – Anti-Government	0.71	0.37	0.48	123
(299) Civil Rights – Other	0.46	0.12	0.19	51
(300) Health – General	0.61	0.6	0.6	824
(301) Health – Health Care Reform	0.49	0.46	0.47	556
(302) Health – Insurance	0.64	0.61	0.62	728
(321) Health – Drug Industry	0.65	0.62	0.64	228
(322) Health – Medical Facilities	0.77	0.69	0.73	422
(323) Health – Insurance Providers	0.77	0.61	0.68	285
(324) Health – Medical Liability	0.78	0.68	0.73	158
(325) Health – Manpower	0.66	0.69	0.67	360
(331) Health – Disease Prevention	0.79	0.78	0.78	783
(332) Health – Infants and Children	0.73	0.73	0.73	242
(333) Health – Mental Health	0.83	0.81	0.82	218
(334) Health – Long-term Care	0.74	0.71	0.72	264
(335) Health – Drug Coverage and Cost	0.78	0.78	0.78	207
(341) Health – Tobacco Abuse	0.84	0.85	0.85	232
(342) Health – Drug and Alcohol Abuse	0.75	0.64	0.69	286
(398) Health – R&D	0.72	0.75	0.73	244
(399) Health – Other	0.5	0.14	0.22	78
(400) Agriculture – General	0.67	0.67	0.67	1116
(401) Agriculture – Trade	0.52	0.5	0.51	325
(402) Agriculture – Subsidies to Farmers	0.71	0.68	0.7	776
(403) Agriculture – Food Inspection & Safety	0.69	0.7	0.69	319
(404) Agriculture – Food Marketing & Promotion	0.64	0.58	0.61	133
(405) Agriculture – Animal and Crop Disease	0.83	0.78	0.81	327
(408) Agriculture – Fisheries & Fishing	0.84	0.74	0.78	207
(498) Agriculture – R&D	0.69	0.75	0.72	111
(499) Agriculture – Other	0.63	0.44	0.52	218
(500) Labor – General	0.67	0.61	0.64	1076
(501) Labor – Worker Safety	0.74	0.71	0.73	259
(502) Labor – Employment Training	0.63	0.61	0.62	777
(503) Labor – Employee Benefits	0.68	0.63	0.65	918
(504) Labor – Labor Unions	0.69	0.66	0.67	742
(505) Labor – Fair Labor Standards	0.59	0.59	0.59	556
(506) Labor – Youth Employment	0.64	0.68	0.66	324
(529) Labor – Migrant and Seasonal	0.63	0.53	0.57	150
(599) Labor – Other	0.7	0.12	0.2	59
(600) Education – General	0.63	0.58	0.61	811
(601) Education – Higher	0.85	0.84	0.84	1836
(602) Education – Elementary & Secondary	0.77	0.8	0.78	2400
(603) Education – Underprivileged	0.51	0.46	0.48	163
(604) Education – Vocational	0.76	0.76	0.76	468
(606) Education – Special	0.73	0.76	0.74	123
(607) Education – Excellence	0.57	0.54	0.55	275
(698) Education – R&D	0.74	0.25	0.37	56
(699) Education – Other	0.72	0.24	0.36	115
(700) Environment – General	0.72	0.73	0.72	979
(701) Environment – Drinking Water	0.7	0.7	0.7	322
(703) Environment – Waste Disposal	0.84	0.82	0.83	367
(704) Environment – Hazardous Waste	0.68	0.75	0.72	208
(705) Environment – Air Pollution	0.78	0.8	0.79	645
(707) Environment – Recycling	0.81	0.89	0.85	63
(708) Environment – Indoor Hazards	0.69	0.67	0.68	49
(709) Environment – Species & Forest	0.76	0.79	0.77	476
(711) Environment – Land and Water Conservation	0.55	0.38	0.45	173
(798) Environment – R&D	0.53	0.2	0.3	44
(799) Environment – Other	0.6	0.06	0.11	51
(800) Energy – General	0.76	0.74	0.75	802
(801) Energy – Nuclear	0.79	0.72	0.75	289
(802) Energy – Electricity	0.72	0.66	0.69	328
(803) Energy – Natural Gas & Oil	0.75	0.77	0.76	646
(805) Energy – Coal	0.73	0.73	0.73	133
(806) Energy – Alternative & Renewable	0.74	0.76	0.75	270
(807) Energy – Conservation	0.65	0.69	0.67	185
(898) Energy – R&D	0.7	0.29	0.41	55
(899) Energy – Other	0	0	0	23
(900) Immigration – Immigration	0.76	0.74	0.75	1411
(999) No Policy Content	0.94	0.97	0.96	89092
(1000) Transportation – General	0.68	0.66	0.67	488
(1001) Transportation – Mass	0.64	0.6	0.62	298
(1002) Transportation – Highways	0.86	0.84	0.85	1715
(1003) Transportation – Air Travel	0.8	0.8	0.8	360
(1005) Transportation – Railroad Travel	0.79	0.8	0.79	539
(1007) Transportation – Maritime	0.79	0.81	0.8	516
(1010) Transportation – Infrastructure	0.56	0.51	0.54	220
(1098) Transportation – R&D	0	0	0	16
(1099) Transportation – Other	0.88	0.58	0.7	52
(1200) Law and Crime – General	0.59	0.56	0.58	714
(1201) Law and Crime – Agencies	0.78	0.79	0.79	1949
(1202) Law and Crime – White Collar Crime	0.56	0.58	0.57	402
(1203) Law and Crime – Illegal Drugs	0.76	0.79	0.78	384
(1204) Law and Crime – Court Administration	0.75	0.74	0.75	1601
(1205) Law and Crime – Prisons	0.79	0.75	0.77	426
(1206) Law and Crime – Juvenile Crime	0.69	0.71	0.7	173
(1207) Law and Crime – Child Abuse	0.77	0.74	0.76	273
(1208) Law and Crime – Family Issues	0.77	0.75	0.76	866
(1210) Law and Crime – Criminal & Civil Code	0.73	0.7	0.71	1480
(1211) Law and Crime – Crime Control	0.61	0.47	0.54	417
(1227) Law and Crime – Police	0.61	0.57	0.59	124
(1299) Law and Crime – Other	0.7	0.38	0.49	165
(1300) Social Welfare – General	0.66	0.62	0.64	1366
(1302) Social Welfare – Low-Income Assistance	0.63	0.63	0.63	589
(1303) Social Welfare – Elderly Assistance	0.74	0.72	0.73	971
(1304) Social Welfare – Disabled Assistance	0.65	0.68	0.66	336
(1305) Social Welfare – Volunteer Associations	0.67	0.65	0.66	422
(1308) Social Welfare – Child Care	0.72	0.77	0.74	351
(1399) Social Welfare – Other	0.5	0.02	0.05	41
(1400) Housing – General	0.69	0.68	0.68	1015
(1401) Housing – Community Development	0.47	0.42	0.44	396
(1403) Housing – Urban Development	0.55	0.54	0.55	329
(1404) Housing – Rural Housing	0	0	0	14
(1405) Housing – Rural Development	0.76	0.66	0.71	486
(1406) Housing – Low-Income Assistance	0.62	0.6	0.61	359
(1407) Housing – Veterans	0.82	0.74	0.78	42
(1408) Housing – Elderly	0.73	0.28	0.4	40
(1409) Housing – Homeless	0.68	0.74	0.71	90
(1498) Housing – R&D	0	0	0	0
(1499) Housing – Other	0.78	0.76	0.77	229
(1500) Domestic Commerce – General	0.62	0.64	0.63	1030
(1501) Domestic Commerce – Banking	0.69	0.65	0.67	745
(1502) Domestic Commerce – Securities & Commodities	0.56	0.57	0.57	282
(1504) Domestic Commerce – Consumer Finance	0.7	0.63	0.66	260
(1505) Domestic Commerce – Insurance Regulation	0.87	0.83	0.85	339
(1507) Domestic Commerce – Bankruptcy	0.66	0.65	0.66	148
(1520) Domestic Commerce – Corporate Management	0.62	0.58	0.6	636
(1521) Domestic Commerce – Small Businesses	0.75	0.73	0.74	651
(1522) Domestic Commerce – Copyrights and Patents	0.83	0.82	0.82	256
(1523) Domestic Commerce – Disaster Relief	0.79	0.75	0.77	421
(1524) Domestic Commerce – Tourism	0.73	0.77	0.75	210
(1525) Domestic Commerce – Consumer Safety	0.75	0.73	0.74	543
(1526) Domestic Commerce – Sports Regulation	0.84	0.86	0.85	794
(1598) Domestic Commerce – R&D	0	0	0	1
(1599) Domestic Commerce – Other	0.84	0.8	0.82	598
(1600) Defense – General	0.68	0.64	0.66	1165
(1602) Defense – Alliances	0.68	0.64	0.66	730
(1603) Defense – Intelligence	0.7	0.66	0.68	289
(1604) Defense – Readiness	0.72	0.73	0.72	564
(1605) Defense – Nuclear Arms	0.81	0.78	0.8	793
(1606) Defense – Military Aid	0.73	0.56	0.63	95
(1608) Defense – Personnel Issues	0.75	0.78	0.77	1284
(1610) Defense – Procurement	0.67	0.71	0.69	211
(1611) Defense – Installations & Land	0.6	0.64	0.62	160
(1612) Defense – Reserve Forces	0.6	0.59	0.59	68
(1614) Defense – Hazardous Waste	0	0	0	12
(1615) Defense – Civil	0.73	0.71	0.72	359
(1616) Defense – Civilian Personnel	0.86	0.28	0.42	65
(1617) Defense – Contractors	0.66	0.53	0.59	79
(1619) Defense – Foreign Operations	0.67	0.64	0.65	866
(1620) Defense – Claims against Military	0.72	0.67	0.69	174
(1698) Defense – R&D	0.63	0.37	0.47	46
(1699) Defense – Other	0.56	0.37	0.44	117
(1700) Technology – General	0.53	0.54	0.53	343
(1701) Technology – Space	0.82	0.86	0.84	129
(1704) Technology – Commercial Use of Space	0.88	0.44	0.58	32
(1705) Technology – Science Transfer	0.63	0.46	0.53	99
(1706) Technology – Telecommunications	0.7	0.74	0.72	345
(1707) Technology – Broadcast	0.9	0.9	0.9	1716
(1708) Technology – Weather Forecasting	0.72	0.77	0.74	62
(1709) Technology – Computers	0.72	0.73	0.73	241
(1798) Technology – R&D	0.74	0.79	0.76	638
(1799) Technology – Other	0.86	0.14	0.24	44
(1800) Foreign Trade – General	0.58	0.6	0.59	375
(1802) Foreign Trade – Trade Agreements	0.75	0.65	0.69	634
(1803) Foreign Trade – Exports	0.72	0.66	0.69	173
(1804) Foreign Trade – Private Investments	0.62	0.6	0.61	196
(1806) Foreign Trade – Competitiveness	0.63	0.62	0.62	329
(1807) Foreign Trade – Tariff & Imports	0.77	0.7	0.73	280
(1808) Foreign Trade – Exchange Rates	0.65	0.39	0.49	79
(1899) Foreign Trade – Other	0	0	0	8
(1900) International Affairs – General	0.71	0.77	0.74	2791
(1901) International Affairs – Foreign Aid	0.69	0.63	0.66	610
(1902) International Affairs – Resources Exploitation	0.62	0.32	0.42	151
(1905) International Affairs – Developing Countries	0.56	0.57	0.56	234
(1906) International Affairs – International Finance	0.64	0.63	0.63	483
(1910) International Affairs – Western Europe	0.68	0.68	0.68	1355
(1921) International Affairs – Specific Country	0.8	0.82	0.81	3664
(1925) International Affairs – Human Rights	0.7	0.7	0.7	505
(1926) International Affairs – Organizations	0.7	0.69	0.7	488
(1927) International Affairs – Terrorism	0.66	0.62	0.64	472
(1929) International Affairs – Diplomats	0.61	0.59	0.6	396
(1999) International Affairs – Other	0.65	0.18	0.28	72
(2000) Government Operations – General	0.67	0.6	0.63	1553
(2001) Government Operations – Intergovernmental Relations	0.66	0.66	0.66	2047
(2002) Government Operations – Bureaucracy	0.61	0.55	0.57	1247
(2003) Government Operations – Postal Service	0.85	0.87	0.86	269
(2004) Government Operations – Employees	0.75	0.74	0.74	1352
(2005) Government Operations – Appointments	0.76	0.78	0.77	540
(2006) Government Operations – Currency	0.71	0.6	0.65	179
(2007) Government Operations – Procurement & Contractors	0.74	0.62	0.67	577
(2008) Government Operations – Property Management	0.73	0.74	0.73	712
(2009) Government Operations – Tax Administration	0.74	0.62	0.67	238
(2010) Government Operations – Scandals	0.66	0.53	0.59	289
(2011) Government Operations – Branch Relations	0.68	0.67	0.68	1919
(2012) Government Operations – Political Campaigns	0.76	0.72	0.74	1821
(2013) Government Operations – Census & Statistics	0.89	0.84	0.87	81
(2014) Government Operations – Capital City	0.85	0.77	0.81	377
(2015) Government Operations – Claims against the Government	0.65	0.7	0.67	746
(2030) Government Operations – National Holidays	0.69	0.66	0.68	296
(2099) Government Operations – Other	0.51	0.35	0.41	323
(2100) Public Lands – General	0.61	0.32	0.42	158
(2101) Public Lands – National Parks	0.77	0.79	0.78	864
(2102) Public Lands – Indigenous Affairs	0.92	0.91	0.92	429
(2103) Public Lands – Public Lands	0.7	0.71	0.71	1137
(2104) Public Lands – Water Resources	0.77	0.75	0.76	760
(2105) Public Lands – Dependencies & Territories	0.66	0.64	0.65	307
(2199) Public Lands – Other	0.56	0.45	0.5	11
(2300) Culture – General	0.75	0.72	0.74	1363

Inference platform

This model is used by the CAP Babel Machine, an open-source and free natural language processing tool, designed to simplify and speed up projects for comparative research.

Cooperation

Model performance can be significantly improved by extending our training sets. We appreciate every submission of CAP-coded corpora (of any domain and language) at poltextlab{at}poltextlab{dot}com or by using the CAP Babel Machine.

Debugging and issues

This architecture uses the sentencepiece tokenizer. In order to run the model before transformers==4.27 you need to install it manually.

Как запустить

На своей GPU в ailist. Цена — по текущим предложениям провайдеров

GPU в ailist

Универсальный рантайм · transformers · text-classification

Нельзя запустить

Доступ к модели ограничен автором; у платформы нет доступа к весам.

Ревизия весов: 509ce32