# Unsupervised POS tagging on WSJ entire corpus (full)

Source: [Unsupervised POS tagging on WSJ entire corpus (full)](<https://www.sota2.com/research/sota/unsupervised-pos-tagging-on-wsj-entire-corpus-full>)

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

Data updated: 2026-02-26T05:48:49.718Z

Evaluation of Unsupervised Part-of-Speech (POS) Tagging performance on the full Wall Street Journal (WSJ) corpus.

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 | Unsupervised POS tagging |
| Dataset (including any stated split) | WSJ entire corpus (full) |
| Metric | M1 Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 13 of 13 |
| 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

- [M1 Score](<https://www.sota2.com/research/sota/unsupervised-pos-tagging-on-wsj-entire-corpus-full.md?metric=M1%20Score>): Higher is better; unit: PERCENT.

- [VM Score](<https://www.sota2.com/research/sota/unsupervised-pos-tagging-on-wsj-entire-corpus-full.md?metric=VM%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | M1 Score | Result date | Report paper |
| --- | --- | --- | --- |
| Invertible Neural Projections | 80.8 | Not recorded | 68fa64bd1e209131cea062e0 |
| Cluster (word-based) | 80.2 | Not recorded | 68fa64bd1e209131cea062e0 |
| Invertible Neural Projections | 79.5 | Not recorded | 68fa64bd1e209131cea062e0 |
| Cluster (token-based) | 79.5 | Not recorded | 68fa64bd1e209131cea062e0 |
| NHMM (+ Conv & LSTM) | 79.1 | Not recorded | 68fa64bd1e209131cea062e0 |
| PYP-HMM | 77.5 | Not recorded | 68fa64bd1e209131cea062e0 |
| Brown (+ proto) | 76.1 | Not recorded | 68fa64bd1e209131cea062e0 |
| Feature HMM | 75.5 | Not recorded | 68fa64bd1e209131cea062e0 |
| Gaussian HMM | 75.4 | Not recorded | 68fa64bd1e209131cea062e0 |
| NHMM (+ Conv) | 74.1 | Not recorded | 68fa64bd1e209131cea062e0 |
| Invertible Neural Projections | 73.2 | Not recorded | 68fa64bd1e209131cea062e0 |
| Discrete HMM | 62.7 | Not recorded | 68fa64bd1e209131cea062e0 |
| NHMM (basic) | 59.8 | Not recorded | 68fa64bd1e209131cea062e0 |

### Invertible Neural Projections

M1 Score: 80.8. Result ID: 698c89ac0b614d12cff16f85.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Implementation](<https://github.com/jxhe/struct-learning-with-flow>)

Recorded evaluation configuration: {"hand-engineered features":"false","Number of layers":"8 layers"}. This record does not assert independent reproduction.

### Cluster (word-based)

M1 Score: 80.2. Result ID: 698c89ac0b614d12cff16f89.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

Recorded evaluation configuration: {"hand-engineered features":"true"}. This record does not assert independent reproduction.

### Invertible Neural Projections

M1 Score: 79.5. Result ID: 698c89ac0b614d12cff16f84.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Implementation](<https://github.com/jxhe/struct-learning-with-flow>)

Recorded evaluation configuration: {"hand-engineered features":"false","Number of layers":"4 layers"}. This record does not assert independent reproduction.

### Cluster (token-based)

M1 Score: 79.5. Result ID: 698c89ac0b614d12cff16f8a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

Recorded evaluation configuration: {"hand-engineered features":"true"}. This record does not assert independent reproduction.

### NHMM (+ Conv & LSTM)

M1 Score: 79.1. Result ID: 698c89ac0b614d12cff16f82.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62d41e209131ce9fb667.md>)

Recorded evaluation configuration: {"hand-engineered features":"false"}. This record does not assert independent reproduction.

### PYP-HMM

M1 Score: 77.5. Result ID: 698c89ac0b614d12cff16f7f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

Recorded evaluation configuration: {"hand-engineered features":"false"}. This record does not assert independent reproduction.

### Brown (+ proto)

M1 Score: 76.1. Result ID: 698c89ac0b614d12cff16f88.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

Recorded evaluation configuration: {"hand-engineered features":"true"}. This record does not assert independent reproduction.

### Feature HMM

M1 Score: 75.5. Result ID: 698c89ac0b614d12cff16f87.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

Recorded evaluation configuration: {"hand-engineered features":"true"}. This record does not assert independent reproduction.

### Gaussian HMM

M1 Score: 75.4. Result ID: 698c89ac0b614d12cff16f83.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa61ca1e209131ce9f5e3e.md>)

Recorded evaluation configuration: {"hand-engineered features":"false"}. This record does not assert independent reproduction.

### NHMM (+ Conv)

M1 Score: 74.1. Result ID: 698c89ac0b614d12cff16f81.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62d41e209131ce9fb667.md>)

Recorded evaluation configuration: {"hand-engineered features":"false"}. This record does not assert independent reproduction.

### Invertible Neural Projections

M1 Score: 73.2. Result ID: 698c89ac0b614d12cff16f86.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Implementation](<https://github.com/jxhe/struct-learning-with-flow>)

Recorded evaluation configuration: {"hand-engineered features":"false","Number of layers":"16 layers"}. This record does not assert independent reproduction.

### Discrete HMM

M1 Score: 62.7. Result ID: 698c89ac0b614d12cff16f7e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

Recorded evaluation configuration: {"hand-engineered features":"false"}. This record does not assert independent reproduction.

### NHMM (basic)

M1 Score: 59.8. Result ID: 698c89ac0b614d12cff16f80.

[Reporting paper](<https://www.sota2.com/research/paper/68fa64bd1e209131cea062e0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa62d41e209131ce9fb667.md>)

Recorded evaluation configuration: {"hand-engineered features":"false"}. This record does not assert independent reproduction.

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