# OCR Verification on Curated dataset of 1,000 PDF pages 1.0 (test)

Source: [OCR Verification on Curated dataset of 1,000 PDF pages 1.0 (test)](<https://www.sota2.com/research/sota/ocr-verification-on-curated-dataset-of-1-000-pdf-pages-1-0-test>)

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

Data updated: 2026-03-18T07:23:10.316Z

Evaluation of OCR Verification performance on the test split of the Curated dataset of 1,000 PDF pages 1.0, typically measured using F1 score.

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 | OCR Verification |
| Dataset (including any stated split) | Curated dataset of 1,000 PDF pages 1.0 (test) |
| Metric | F1 Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 30 of 30 |
| 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

- [F1 Score](<https://www.sota2.com/research/sota/ocr-verification-on-curated-dataset-of-1-000-pdf-pages-1-0-test.md?metric=F1%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | F1 Score | Result date | Report paper |
| --- | --- | --- | --- |
| VLM-as-Judge | 72.27 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 72.12 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 70.88 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 70.05 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 69.57 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 67.42 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 66.13 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 65.44 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 57.29 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 54.74 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 52.83 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 51.3 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 51 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 48.21 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 48 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 40.19 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 40.14 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 40 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 39.8 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 39.56 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 38.67 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 36.1 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 28.47 | Not recorded | 68fa78051e209131cea765d8 |
| CE | 27.34 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 23.01 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 21.47 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 19.82 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 18.42 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 11.43 | Not recorded | 68fa78051e209131cea765d8 |
| VLM-as-Judge | 0 | Not recorded | 68fa78051e209131cea765d8 |

### VLM-as-Judge

F1 Score: 72.27. Result ID: 69ba52c747f34ce924a3caf4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.9-1.0"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 72.12. Result ID: 69ba52c747f34ce924a3cae0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.9-1.0"}. This record does not assert independent reproduction.

### CE

F1 Score: 70.88. Result ID: 69ba52c747f34ce924a3caeb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.9-1.0"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 70.05. Result ID: 69ba52c747f34ce924a3caea.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.9-1.0"}. This record does not assert independent reproduction.

### CE

F1 Score: 69.57. Result ID: 69ba52c747f34ce924a3cafb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.0-0.3"}. This record does not assert independent reproduction.

### CE

F1 Score: 67.42. Result ID: 69ba52c747f34ce924a3caf1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.0-0.3"}. This record does not assert independent reproduction.

### CE

F1 Score: 66.13. Result ID: 69ba52c747f34ce924a3caf5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.9-1.0"}. This record does not assert independent reproduction.

### CE

F1 Score: 65.44. Result ID: 69ba52c747f34ce924a3cae7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.0-0.3"}. This record does not assert independent reproduction.

### CE

F1 Score: 57.29. Result ID: 69ba52c747f34ce924a3cae1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.9-1.0"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 54.74. Result ID: 69ba52c747f34ce924a3caf0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.0-0.3"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 52.83. Result ID: 69ba52c747f34ce924a3cafa.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.0-0.3"}. This record does not assert independent reproduction.

### CE

F1 Score: 51.3. Result ID: 69ba52c747f34ce924a3caf3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"Overall"}. This record does not assert independent reproduction.

### CE

F1 Score: 51. Result ID: 69ba52c747f34ce924a3cafd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"Overall"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 48.21. Result ID: 69ba52c747f34ce924a3cae6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.0-0.3"}. This record does not assert independent reproduction.

### CE

F1 Score: 48. Result ID: 69ba52c747f34ce924a3cae9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"Overall"}. This record does not assert independent reproduction.

### CE

F1 Score: 40.19. Result ID: 69ba52c747f34ce924a3cae3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.7-0.8"}. This record does not assert independent reproduction.

### CE

F1 Score: 40.14. Result ID: 69ba52c747f34ce924a3caf7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.7-0.8"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 40. Result ID: 69ba52c747f34ce924a3cae8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"Overall"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 39.8. Result ID: 69ba52c747f34ce924a3cafc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"Overall"}. This record does not assert independent reproduction.

### CE

F1 Score: 39.56. Result ID: 69ba52c747f34ce924a3caed.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.7-0.8"}. This record does not assert independent reproduction.

### CE

F1 Score: 38.67. Result ID: 69ba52c747f34ce924a3cae5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.4-0.6"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 36.1. Result ID: 69ba52c747f34ce924a3caf2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"Overall"}. This record does not assert independent reproduction.

### CE

F1 Score: 28.47. Result ID: 69ba52c747f34ce924a3caf9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.4-0.6"}. This record does not assert independent reproduction.

### CE

F1 Score: 27.34. Result ID: 69ba52c747f34ce924a3caef.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.4-0.6"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 23.01. Result ID: 69ba52c747f34ce924a3caf6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.7-0.8"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 21.47. Result ID: 69ba52c747f34ce924a3cae4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.4-0.6"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 19.82. Result ID: 69ba52c747f34ce924a3caec.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.7-0.8"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 18.42. Result ID: 69ba52c747f34ce924a3cae2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"GPT4o","Human Score Band":"0.7-0.8"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 11.43. Result ID: 69ba52c747f34ce924a3caf8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-72B","Human Score Band":"0.4-0.6"}. This record does not assert independent reproduction.

### VLM-as-Judge

F1 Score: 0. Result ID: 69ba52c747f34ce924a3caee.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78051e209131cea765d8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Reference VLM":"Qwen2-VL-7B","Human Score Band":"0.4-0.6"}. This record does not assert independent reproduction.

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