# Part-of-Speech Tagging on POS (test)

Source: [Part-of-Speech Tagging on POS (test)](<https://www.sota2.com/research/sota/part-of-speech-tagging-on-pos-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:28:53.808Z

Evaluation of Part-of-Speech (POS) Tagging performance on the test split of the POS dataset, typically measured using Accuracy.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Part-of-Speech Tagging |
| Dataset (including any stated split) | POS (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 33 of 33 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Accuracy](<https://www.sota2.com/research/sota/part-of-speech-tagging-on-pos-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| ELMo + Multi-task | 97.79 | Not recorded | 68fa64da1e209131cea06bba |
| ELMo + Multi-task | 97.79 | Not recorded | 68fa64da1e209131cea06bba |
| CVT + Multi-task | 97.76 | Not recorded | 68fa64da1e209131cea06bba |
| CVT + Multi-Task | 97.76 | Not recorded | 68fa64da1e209131cea06bba |
| CVT + Multi-task + Large | 97.74 | Not recorded | 68fa64da1e209131cea06bba |
| CVT + Multi-Task + Large | 97.74 | Not recorded | 68fa64da1e209131cea06bba |
| ELMo (our implementation) | 97.72 | Not recorded | 68fa64da1e209131cea06bba |
| ELMo | 97.72 | Not recorded | 68fa64da1e209131cea06bba |
| CVT | 97.7 | Not recorded | 68fa64da1e209131cea06bba |
| CVT | 97.7 | Not recorded | 68fa64da1e209131cea06bba |
| Word Dropout | 97.66 | Not recorded | 68fa64da1e209131cea06bba |
| Word Dropout | 97.66 | Not recorded | 68fa64da1e209131cea06bba |
| Virtual Adversarial Training | 97.64 | Not recorded | 68fa64da1e209131cea06bba |
| Virtual Adversarial Training | 97.64 | Not recorded | 68fa64da1e209131cea06bba |
| Supervised | 97.6 | Not recorded | 68fa64da1e209131cea06bba |
| Supervised | 97.6 | Not recorded | 68fa64da1e209131cea06bba |
| JMT | 97.55 | Not recorded | 68fa64da1e209131cea06bba |
| LSTM-CNN-CRF | 97.55 | Not recorded | 68fa64da1e209131cea06bba |
| JMT\* | 97.55 | Not recorded | 68fa64da1e209131cea06bba |
| Shortcut LSTM | 97.53 | Not recorded | 68fa64da1e209131cea06bba |
| Shortcut LSTM | 97.53 | Not recorded | 68fa64da1e209131cea06bba |
| LM-LSTM-CNN-CRF | 97.53 | Not recorded | 68fa64da1e209131cea06bba |
| ZEN (P) | 96.64 | Not recorded | 68fa669b1e209131cea115e1 |
| BERT (P) | 96.1 | Not recorded | 68fa669b1e209131cea115e1 |
| ZEN (R) | 95.82 | Not recorded | 68fa669b1e209131cea115e1 |
| BERT (R) | 95.43 | Not recorded | 68fa669b1e209131cea115e1 |
| Ref. | 92.28 | Not recorded | 68fa6f6c1e209131cea45774 |
| TAPIR-Trf | 91.49 | Not recorded | 68fa6f6c1e209131cea45774 |
| TAPIR-Trf | 91.35 | Not recorded | 68fa6f6c1e209131cea45774 |
| TAPIR-Trf | 91.32 | Not recorded | 68fa6f6c1e209131cea45774 |
| TAPIR-LT | 90.9 | Not recorded | 68fa6f6c1e209131cea45774 |
| TAPIR-LT | 90.83 | Not recorded | 68fa6f6c1e209131cea45774 |
| TAPIR-LT | 90.65 | Not recorded | 68fa6f6c1e209131cea45774 |

### ELMo + Multi-task

Accuracy: 97.79. Result ID: 698c8f600b614d12cff19674.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64271e209131cea02f00.md>)

[Implementation](<http://allennlp.org/elmo>)

Recorded evaluation configuration: {"Semi-supervised":true,"Multi-task":true}. This record does not assert independent reproduction.

### ELMo + Multi-task

Accuracy: 97.79. Result ID: 698c8f600b614d12cff196be.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64271e209131cea02f00.md>)

[Implementation](<http://allennlp.org/elmo>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Multi-task","Model Scale":"Standard"}. This record does not assert independent reproduction.

### CVT + Multi-task

Accuracy: 97.76. Result ID: 698c8f600b614d12cff19676.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Implementation](<https://github.com/tensorflow/models/tree/master/research/cvt_text>)

Recorded evaluation configuration: {"Semi-supervised":true,"Multi-task":true}. This record does not assert independent reproduction.

### CVT + Multi-Task

Accuracy: 97.76. Result ID: 698c8f600b614d12cff196c0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Implementation](<https://github.com/tensorflow/models/tree/master/research/cvt_text>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Multi-task","Model Scale":"Standard"}. This record does not assert independent reproduction.

### CVT + Multi-task + Large

Accuracy: 97.74. Result ID: 698c8f600b614d12cff19677.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Implementation](<https://github.com/tensorflow/models/tree/master/research/cvt_text>)

Recorded evaluation configuration: {"Semi-supervised":true,"Multi-task":true,"Model Size":"Large"}. This record does not assert independent reproduction.

### CVT + Multi-Task + Large

Accuracy: 97.74. Result ID: 698c8f600b614d12cff196c1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Implementation](<https://github.com/tensorflow/models/tree/master/research/cvt_text>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Multi-task","Model Scale":"Large"}. This record does not assert independent reproduction.

### ELMo (our implementation)

Accuracy: 97.72. Result ID: 698c8f600b614d12cff19673.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64271e209131cea02f00.md>)

[Implementation](<http://allennlp.org/elmo>)

Recorded evaluation configuration: {"Semi-supervised":true,"Multi-task":false}. This record does not assert independent reproduction.

### ELMo

Accuracy: 97.72. Result ID: 698c8f600b614d12cff196bd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64271e209131cea02f00.md>)

[Implementation](<http://allennlp.org/elmo>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Single-task","Model Scale":"Standard"}. This record does not assert independent reproduction.

### CVT

Accuracy: 97.7. Result ID: 698c8f600b614d12cff19675.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Implementation](<https://github.com/tensorflow/models/tree/master/research/cvt_text>)

Recorded evaluation configuration: {"Semi-supervised":true,"Multi-task":false}. This record does not assert independent reproduction.

### CVT

Accuracy: 97.7. Result ID: 698c8f600b614d12cff196bf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Implementation](<https://github.com/tensorflow/models/tree/master/research/cvt_text>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Single-task","Model Scale":"Standard"}. This record does not assert independent reproduction.

### Word Dropout

Accuracy: 97.66. Result ID: 698c8f600b614d12cff19672.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

Recorded evaluation configuration: {"Semi-supervised":true,"Multi-task":false}. This record does not assert independent reproduction.

### Word Dropout

Accuracy: 97.66. Result ID: 698c8f600b614d12cff196bc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Single-task","Model Scale":"Standard"}. This record does not assert independent reproduction.

### Virtual Adversarial Training

Accuracy: 97.64. Result ID: 698c8f600b614d12cff19671.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62891e209131ce9f9fa7.md>)

Recorded evaluation configuration: {"Semi-supervised":true,"Multi-task":false}. This record does not assert independent reproduction.

### Virtual Adversarial Training

Accuracy: 97.64. Result ID: 698c8f600b614d12cff196bb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62891e209131ce9f9fa7.md>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Single-task","Model Scale":"Standard"}. This record does not assert independent reproduction.

### Supervised

Accuracy: 97.6. Result ID: 698c8f600b614d12cff19670.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

Recorded evaluation configuration: {"Semi-supervised":false,"Multi-task":false}. This record does not assert independent reproduction.

### Supervised

Accuracy: 97.6. Result ID: 698c8f600b614d12cff196ba.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

Recorded evaluation configuration: {"Supervision Type":"Supervised","Learning Paradigm":"Single-task","Model Scale":"Standard"}. This record does not assert independent reproduction.

### JMT

Accuracy: 97.55. Result ID: 698c8f600b614d12cff1966f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63801e209131ce9fef77.md>)

[Implementation](<https://github.com/hassyGo/charNgram2vec>)

Recorded evaluation configuration: {"Semi-supervised":false,"Multi-task":true}. This record does not assert independent reproduction.

### LSTM-CNN-CRF

Accuracy: 97.55. Result ID: 698c8f600b614d12cff196b6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62861e209131ce9f9e7e.md>)

Recorded evaluation configuration: {"Supervision Type":"Supervised"}. This record does not assert independent reproduction.

### JMT\*

Accuracy: 97.55. Result ID: 698c8f600b614d12cff196b8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63801e209131ce9fef77.md>)

[Implementation](<https://github.com/hassyGo/charNgram2vec>)

Recorded evaluation configuration: {"Supervision Type":"Semi-supervised","Learning Paradigm":"Multi-task"}. This record does not assert independent reproduction.

### Shortcut LSTM

Accuracy: 97.53. Result ID: 698c8f600b614d12cff1966e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63041e209131ce9fc5db.md>)

Recorded evaluation configuration: {"Semi-supervised":false,"Multi-task":false}. This record does not assert independent reproduction.

### Shortcut LSTM

Accuracy: 97.53. Result ID: 698c8f600b614d12cff196b7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63041e209131ce9fc5db.md>)

Recorded evaluation configuration: {"Supervision Type":"Supervised"}. This record does not assert independent reproduction.

### LM-LSTM-CNN-CRF

Accuracy: 97.53. Result ID: 698c8f600b614d12cff196b9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64da1e209131cea06bba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa63d01e209131cea00da2.md>)

[Implementation](<https://github.com/LiyuanLucasLiu/LM-LSTM-CRF>)

Recorded evaluation configuration: {"Supervision Type":"Supervised"}. This record does not assert independent reproduction.

### ZEN (P)

Accuracy: 96.64. Result ID: 698cfc158b517eb4331e2a17.

[Reporting paper](<https://www.sota2.com/research/paper/68fa669b1e209131cea115e1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa669b1e209131cea115e1.md>)

[Implementation](<https://github.com/sinovation/ZEN>)

Recorded evaluation configuration: {"Initialization":"Pre-trained","Backbone Scale":"Base"}. This record does not assert independent reproduction.

### BERT (P)

Accuracy: 96.1. Result ID: 698cfc158b517eb4331e2a15.

[Reporting paper](<https://www.sota2.com/research/paper/68fa669b1e209131cea115e1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65c31e209131cea0c4ed.md>)

[Implementation](<https://github.com/google-research/bert>)

Recorded evaluation configuration: {"Initialization":"Pre-trained","Backbone Scale":"Base"}. This record does not assert independent reproduction.

### ZEN (R)

Accuracy: 95.82. Result ID: 698cfc158b517eb4331e2a16.

[Reporting paper](<https://www.sota2.com/research/paper/68fa669b1e209131cea115e1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa669b1e209131cea115e1.md>)

[Implementation](<https://github.com/sinovation/ZEN>)

Recorded evaluation configuration: {"Initialization":"Random","Backbone Scale":"Base"}. This record does not assert independent reproduction.

### BERT (R)

Accuracy: 95.43. Result ID: 698cfc158b517eb4331e2a14.

[Reporting paper](<https://www.sota2.com/research/paper/68fa669b1e209131cea115e1.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa65c31e209131cea0c4ed.md>)

[Implementation](<https://github.com/google-research/bert>)

Recorded evaluation configuration: {"Initialization":"Random","Backbone Scale":"Base"}. This record does not assert independent reproduction.

### Ref.

Accuracy: 92.28. Result ID: 6990bca93a6c31484902de0a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72ab1e209131cea57bd2.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### TAPIR-Trf

Accuracy: 91.49. Result ID: 6990bca93a6c31484902de0d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Implementation](<https://github.com/pkhdipraja/tapir>)

Recorded evaluation configuration: {"delay":2}. This record does not assert independent reproduction.

### TAPIR-Trf

Accuracy: 91.35. Result ID: 6990bca93a6c31484902de0c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Implementation](<https://github.com/pkhdipraja/tapir>)

Recorded evaluation configuration: {"delay":1}. This record does not assert independent reproduction.

### TAPIR-Trf

Accuracy: 91.32. Result ID: 6990bca93a6c31484902de0b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Implementation](<https://github.com/pkhdipraja/tapir>)

Recorded evaluation configuration: {"delay":0}. This record does not assert independent reproduction.

### TAPIR-LT

Accuracy: 90.9. Result ID: 6990bca93a6c31484902de0e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Implementation](<https://github.com/pkhdipraja/tapir>)

Recorded evaluation configuration: {"delay":0}. This record does not assert independent reproduction.

### TAPIR-LT

Accuracy: 90.83. Result ID: 6990bca93a6c31484902de0f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Implementation](<https://github.com/pkhdipraja/tapir>)

Recorded evaluation configuration: {"delay":1}. This record does not assert independent reproduction.

### TAPIR-LT

Accuracy: 90.65. Result ID: 6990bca93a6c31484902de10.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6c1e209131cea45774.md>)

[Implementation](<https://github.com/pkhdipraja/tapir>)

Recorded evaluation configuration: {"delay":2}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.