UD Latin ITTB
Models analyze medieval Latin prose by Thomas Aquinas. They identify the grammatical form of each word and its relationship to other words.
Bamman and Burns evaluated Latin BERT on part of speech tagging. Kupari and colleagues also scored models on the treebank's grammatical dependencies.
- Made by
- Index Thomisticus Treebank team
- Released
- November 2015
- Kind
- Reading in many languages
- Level
- University
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- All public
- Measures
- Capability
- Subject
- Languages and literature
- Language
- Latin
- Interaction
- Single turn
- Items from
- Real work and use, Another benchmark
- Refresh
- New versions replace it
- Run by
- Anyone
- Based on
- A version of another benchmark
HIT-SCIR (Harbin) tops the leaderboard with 87.1%
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.
- 87.1 HIT-SCIR (Harbin)
- 86.5 TurkuNLP (Turku)
- 86.3 NLP-Cube (București)
- 85.9 Stanford (Stanford)
- 85.5 ICS PAS (Warszawa)
- 85.2 UDPipe Future (Praha)
- 85.2 CEA LIST (Paris)
- 84.7 LATTICE (Paris)
- 84.6 ParisNLP (Paris)
- 84.0 LeisureX (Shanghai)
- 83.9 SLT-Interactions (Bengaluru)
- 83.1 AntNLP (Shanghai)
- 83.0 Uppsala (Uppsala)
- 82.9 iParse (Pittsburgh)
- 82.4 IBM NY (Yorktown Heights)
- 79.1 Fudan (Shanghai)
- 78.1 UniMelb (Melbourne)
- 76.3 BOUN (İstanbul)
- 76.0 BASELINE UDPipe 1.2 (Praha)
- 75.8 KParse (İstanbul)
- 75.7 Phoenix (Shanghai)
- 73.8 HUJI (Yerushalayim)
- 73.5 CUNI x-ling (Praha)
- 63.2 ONLP lab (Ra'anana)
- 60.0 ArmParser (Yerevan)
- 0.0 SParse (İstanbul)
- 1 HIT-SCIR (Harbin) software1-P 87.1%
- 2 TurkuNLP (Turku) software1-P 86.5%
- 3 NLP-Cube (București) software1-P 86.3%
- 4 Stanford (Stanford) software2 85.9%
- 5 ICS PAS (Warszawa) software1-P 85.5%
- 6 UDPipe Future (Praha) software1-P 85.2%
- 7 CEA LIST (Paris) software1-P 85.2%
- 8 LATTICE (Paris) software1-P 84.7%
- 9 ParisNLP (Paris) software1-P 84.6%
- 10 LeisureX (Shanghai) software1-P 84%
- 11 SLT-Interactions (Bengaluru) software2-P 83.9%
- 12 AntNLP (Shanghai) software1-P 83.1%
- 13 Uppsala (Uppsala) software1-P 83%
- 14 iParse (Pittsburgh) software1-P 82.9%
- 15 IBM NY (Yorktown Heights) software1-P 82.4%
- 16 Fudan (Shanghai) software5-P 79.1%
- 17 UniMelb (Melbourne) software1-P 78.1%
- 18 BOUN (İstanbul) software2-P 76.3%
- 19 BASELINE UDPipe 1.2 (Praha) software1-P 76%
- 20 KParse (İstanbul) software1-P 75.8%
- 21 Phoenix (Shanghai) software1-P 75.7%
- 22 HUJI (Yerushalayim) software1-P 73.8%
- 23 CUNI x-ling (Praha) software1-P 73.5%
- 24 ONLP lab (Ra'anana) software3-P 63.2%
- 25 ArmParser (Yerevan) software1-P 60%
- 26 SParse (İstanbul) software1-P 0%
Source: Official UD Latin ITTB leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | HIT-SCIR (Harbin) | 87.1% | software1-P | Official leaderboard | |
| 2 | TurkuNLP (Turku) | 86.5% | software1-P | Official leaderboard | |
| 3 | NLP-Cube (București) | 86.3% | software1-P | Official leaderboard | |
| 4 | Stanford (Stanford) | 85.9% | software2 | Official leaderboard | |
| 5 | ICS PAS (Warszawa) | 85.5% | software1-P | Official leaderboard | |
| 6 | UDPipe Future (Praha) | 85.2% | software1-P | Official leaderboard | |
| 7 | CEA LIST (Paris) | 85.2% | software1-P | Official leaderboard | |
| 8 | LATTICE (Paris) | 84.7% | software1-P | Official leaderboard | |
| 9 | ParisNLP (Paris) | 84.6% | software1-P | Official leaderboard | |
| 10 | LeisureX (Shanghai) | 84% | software1-P | Official leaderboard | |
| 11 | SLT-Interactions (Bengaluru) | 83.9% | software2-P | Official leaderboard | |
| 12 | AntNLP (Shanghai) | 83.1% | software1-P | Official leaderboard | |
| 13 | Uppsala (Uppsala) | 83% | software1-P | Official leaderboard | |
| 14 | iParse (Pittsburgh) | 82.9% | software1-P | Official leaderboard | |
| 15 | IBM NY (Yorktown Heights) | 82.4% | software1-P | Official leaderboard | |
| 16 | Fudan (Shanghai) | 79.1% | software5-P | Official leaderboard | |
| 17 | UniMelb (Melbourne) | 78.1% | software1-P | Official leaderboard | |
| 18 | BOUN (İstanbul) | 76.3% | software2-P | Official leaderboard | |
| 19 | BASELINE UDPipe 1.2 (Praha) | 76% | software1-P | Official leaderboard | |
| 20 | KParse (İstanbul) | 75.8% | software1-P | Official leaderboard | |
| 21 | Phoenix (Shanghai) | 75.7% | software1-P | Official leaderboard | |
| 22 | HUJI (Yerushalayim) | 73.8% | software1-P | Official leaderboard | |
| 23 | CUNI x-ling (Praha) | 73.5% | software1-P | Official leaderboard | |
| 24 | ONLP lab (Ra'anana) | 63.2% | software3-P | Official leaderboard | |
| 25 | ArmParser (Yerevan) | 60% | software1-P | Official leaderboard | |
| 26 | SParse (İstanbul) | 0% | software1-P | Official leaderboard |
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Multilingual
- MMMLU In use MMLU professionally translated into 14 languages. Multiple-choice accuracy. Claude Mythos Preview 92.7%
- WMT24++ In use English-to-55-language translation of WMT24 test data. Scored with automatic metrics such as MetricX or xCOMET.
K-EXAONE 90.5%
- MMLU-ProX In use MMLU-Pro translated into up to 29 languages with expert review. Multiple-choice accuracy per language.
Qwen3.7-Max 87%
- CLIcK In use Models answer 1,995 multiple-choice questions on Korean language and culture. Questions come from official Korean exams and textbooks in eleven categories.
Solar Open 2 90.7%
- Global MMLU Lite 3.0 In use Models answer 400 human-translated MMLU questions per language in 23 languages. Version 3.0 adds Czech, Hungarian, Oriya, Slovak and Tajik and updates Italian. Accuracy. Claude Opus 5 95.7%
- MMMB In use Models answer multiple-choice questions about images in six languages across 15 categories. The set has about 12,000 questions. LFM2.5-VL-3B 83%
Data audit
All 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (26 entries).
- Pass Models scored 26 entries on its leaderboard, none a model we track.
- Note Newest score No scores yet.
- Note Frontier models No models scored yet.
- Note Headroom No scores yet.
Leaderboard import 3 pass
- Pass Importer We read the board on universaldependencies.org: 26 entries at the last read, none a model we track.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.