# Radiology Report Generation on MIMIC-CXR v1.5 (Vicuna-7B)

Source: [Radiology Report Generation on MIMIC-CXR v1.5 (Vicuna-7B)](<https://www.sota2.com/research/sota/radiology-report-generation-on-mimic-cxr-v1-5-vicuna-7b>)

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

Data updated: 2026-02-26T06:00:44.088Z

Evaluation of models generating radiology reports using the MIMIC-CXR Findings generation dataset (v1.5), measured using ROUGE-L, BLEU-4, RG\_ER, and Macro-F1-14.

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 | Radiology report generation |
| Dataset (including any stated split) | MIMIC-CXR Findings generation v1.5 (Vicuna-7B) |
| Metric | ROUGE-L |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 12 of 12 |
| 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

- [ROUGE-L](<https://www.sota2.com/research/sota/radiology-report-generation-on-mimic-cxr-v1-5-vicuna-7b.md?metric=ROUGE-L>): Higher is better; unit: PERCENT.

- [BLEU-4](<https://www.sota2.com/research/sota/radiology-report-generation-on-mimic-cxr-v1-5-vicuna-7b.md?metric=BLEU-4>): Higher is better; unit: PERCENT.

- [RG\_ER](<https://www.sota2.com/research/sota/radiology-report-generation-on-mimic-cxr-v1-5-vicuna-7b.md?metric=RG_ER>): Lower is better; unit: PERCENT.

- [Macro-F1-14](<https://www.sota2.com/research/sota/radiology-report-generation-on-mimic-cxr-v1-5-vicuna-7b.md?metric=Macro-F1-14>): Higher is better; unit: PERCENT.

## Recorded results

| Method | ROUGE-L | Result date | Report paper |
| --- | --- | --- | --- |
| RadJEPA | 26.1 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| I-JEPA | 25.6 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| RadJEPAcontrol | 25.5 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| RAD-DINO | 25.1 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| RAD-DINO | 24.6 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| RAD-DINOcontrol | 24.2 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| BioViL | 23.5 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| CLIP@336 | 23.3 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| CheXzero | 23.2 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| BiomedCLIP | 23.1 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| CLIP@224 | 23 | Not recorded | 6973007cb2be0ed85a78c9d3 |
| DINO-v2 | 22.7 | Not recorded | 6973007cb2be0ed85a78c9d3 |

### RadJEPA

ROUGE-L: 26.1. Result ID: 69979a185698665ec4efde91.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

Recorded evaluation configuration: {"Input resolution":"224x224","# Tokens":1369}. This record does not assert independent reproduction.

### I-JEPA

ROUGE-L: 25.6. Result ID: 69979a185698665ec4efde8f.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f061e209131cea43581.md>)

Recorded evaluation configuration: {"Input resolution":"224x224","# Tokens":1369}. This record does not assert independent reproduction.

### RadJEPAcontrol

ROUGE-L: 25.5. Result ID: 69979a185698665ec4efde90.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

Recorded evaluation configuration: {"Input resolution":"224x224","# Tokens":1369}. This record does not assert independent reproduction.

### RAD-DINO

ROUGE-L: 25.1. Result ID: 69979a185698665ec4efde8e.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76e61e209131cea6fc77.md>)

Recorded evaluation configuration: {"Input resolution":"518x518","# Tokens":1369,"reproduced":true}. This record does not assert independent reproduction.

### RAD-DINO

ROUGE-L: 24.6. Result ID: 69979a185698665ec4efde8d.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76e61e209131cea6fc77.md>)

Recorded evaluation configuration: {"Input resolution":"518x518","# Tokens":1369}. This record does not assert independent reproduction.

### RAD-DINOcontrol

ROUGE-L: 24.2. Result ID: 69979a185698665ec4efde8c.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

Recorded evaluation configuration: {"Input resolution":"518x518","# Tokens":1369}. This record does not assert independent reproduction.

### BioViL

ROUGE-L: 23.5. Result ID: 69979a185698665ec4efde8b.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa45541e209131ce917ce9.md>)

Recorded evaluation configuration: {"Input resolution":"512x512","# Tokens":196}. This record does not assert independent reproduction.

### CLIP@336

ROUGE-L: 23.3. Result ID: 69979a185698665ec4efde87.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: {"Input resolution":"316x316","# Tokens":576}. This record does not assert independent reproduction.

### CheXzero

ROUGE-L: 23.2. Result ID: 69979a185698665ec4efde8a.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa425b1e209131ce901aad.md>)

Recorded evaluation configuration: {"Input resolution":"224x224","# Tokens":49}. This record does not assert independent reproduction.

### BiomedCLIP

ROUGE-L: 23.1. Result ID: 69979a185698665ec4efde89.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa768f1e209131cea6db5b.md>)

Recorded evaluation configuration: {"Input resolution":"224x224","# Tokens":256}. This record does not assert independent reproduction.

### CLIP@224

ROUGE-L: 23. Result ID: 69979a185698665ec4efde86.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69581e209131cea21ad0.md>)

[Implementation](<https://github.com/OpenAI/CLIP>)

Recorded evaluation configuration: {"Input resolution":"224x224","# Tokens":256}. This record does not assert independent reproduction.

### DINO-v2

ROUGE-L: 22.7. Result ID: 69979a185698665ec4efde88.

[Reporting paper](<https://www.sota2.com/research/paper/6973007cb2be0ed85a78c9d3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72181e209131cea546c3.md>)

[Implementation](<https://github.com/facebookresearch/dinov2>)

Recorded evaluation configuration: {"Input resolution":"518x518","# Tokens":1369}. This record does not assert independent reproduction.

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