# Handwriting Recognition on CVL Modern Clean Handwriting (test)

Source: [Handwriting Recognition on CVL Modern Clean Handwriting (test)](<https://www.sota2.com/research/sota/handwriting-recognition-on-cvl-modern-clean-handwriting-test>)

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

Data updated: 2026-03-31T10:43:04.628Z

Evaluation of Handwriting Recognition performance on the test split of the CVL Modern Clean Handwriting dataset, measured by Word Accuracy and Character Error Rate (CER).

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 | Handwriting Recognition |
| Dataset (including any stated split) | CVL Modern Clean Handwriting (test) |
| Metric | Word Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [Word Accuracy](<https://www.sota2.com/research/sota/handwriting-recognition-on-cvl-modern-clean-handwriting-test.md?metric=Word%20Accuracy>): Higher is better; unit: PERCENT.

- [Character Error Rate (CER)](<https://www.sota2.com/research/sota/handwriting-recognition-on-cvl-modern-clean-handwriting-test.md?metric=Character%20Error%20Rate%20(CER)>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Word Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| MGP-STR | 82.3 | Not recorded | 69cb553b0636e5bf4afca0f8 |
| T5-Base | 78.1 | Not recorded | 69cb553b0636e5bf4afca0f8 |
| BART-Base | 77.82 | Not recorded | 69cb553b0636e5bf4afca0f8 |
| ByT5-Base | 76.45 | Not recorded | 69cb553b0636e5bf4afca0f8 |
| TrOCR-Base | 74.5 | Not recorded | 69cb553b0636e5bf4afca0f8 |
| Detector Baseline | 72.84 | Not recorded | 69cb553b0636e5bf4afca0f8 |

### MGP-STR

Word Accuracy: 82.3. Result ID: 69cba52547f34ce924b4a760.

[Reporting paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

Recorded evaluation configuration: {"Architecture":"Multi-Granularity ViT","Params":"148M","Training Data":"100M+ Synthetic","Adaptation Cost":"100+ GPU hrs"}. This record does not assert independent reproduction.

### T5-Base

Word Accuracy: 78.1. Result ID: 69cba52547f34ce924b4a761.

[Reporting paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

Recorded evaluation configuration: {"Architecture":"Linguistic Adapter","Params":"220M","Training Data":"1.5M Synthetic","Adaptation Cost":"3 + 0.5 GPU hrs"}. This record does not assert independent reproduction.

### BART-Base

Word Accuracy: 77.82. Result ID: 69cba52547f34ce924b4a762.

[Reporting paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

Recorded evaluation configuration: {"Architecture":"Linguistic Adapter","Params":"140M","Training Data":"1.5M Synthetic","Adaptation Cost":"3 + 0.5 GPU hrs"}. This record does not assert independent reproduction.

### ByT5-Base

Word Accuracy: 76.45. Result ID: 69cba52547f34ce924b4a763.

[Reporting paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

Recorded evaluation configuration: {"Architecture":"Linguistic Adapter","Params":"582M","Training Data":"1.5M Synthetic","Adaptation Cost":"3 + 1.0 GPU hrs"}. This record does not assert independent reproduction.

### TrOCR-Base

Word Accuracy: 74.5. Result ID: 69cba52547f34ce924b4a764.

[Reporting paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d5a1e209131cea3964c.md>)

[Implementation](<https://aka.ms/trocr>)

Recorded evaluation configuration: {"Architecture":"End-to-End Transform.","Params":"334M","Training Data":"684M Synth Lines","Adaptation Cost":"200+ GPU hrs"}. This record does not assert independent reproduction.

### Detector Baseline

Word Accuracy: 72.84. Result ID: 69cba52547f34ce924b4a765.

[Reporting paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb553b0636e5bf4afca0f8.md>)

Recorded evaluation configuration: {"Architecture":"Object Detector Only","Params":"40M","Training Data":"1M Synthetic","Adaptation Cost":"3 GPU hrs"}. This record does not assert independent reproduction.

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