# Large Language Model Watermarking on Mistral-7B-Instruct (test)

Source: [Large Language Model Watermarking on Mistral-7B-Instruct (test)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test>)

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

Data updated: 2026-02-26T05:57:58.031Z

Evaluation of watermarking effectiveness and potential impact on generation quality (measured by PPL) for text generated by the Mistral-7B-Instruct model on its designated test set, utilizing various detection metrics like AUC and WR.

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 | Large Language Model Watermarking |
| Dataset (including any stated split) | Mistral-7B-Instruct (test) |
| Metric | Perplexity (PPL) |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 34 of 34 |
| 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

- [Perplexity (PPL)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=Perplexity%20(PPL)>): Lower is better; unit: SCALAR.

- [Watermark Rate (WR)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=Watermark%20Rate%20(WR)>): Higher is better; unit: PERCENT.

- [Watermark Rate (WR, k=3)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=Watermark%20Rate%20(WR%2C%20k%3D3)>): Higher is better; unit: PERCENT.

- [AUC](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=AUC>): Higher is better; unit: PERCENT.

- [pAUC](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=pAUC>): Higher is better; unit: PERCENT.

- [AUC (C)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=AUC%20(C)>): Higher is better; unit: PERCENT.

- [pAUC (C)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=pAUC%20(C)>): Higher is better; unit: PERCENT.

- [AUC (P)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=AUC%20(P)>): Higher is better; unit: PERCENT.

- [pAUC (P)](<https://www.sota2.com/research/sota/large-language-model-watermarking-on-mistral-7b-instruct-test.md?metric=pAUC%20(P)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Perplexity (PPL) | Result date | Report paper |
| --- | --- | --- | --- |
| Greedy Decoding | 1.37 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 2.61 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 2.63 | Not recorded | 68fa77c11e209131cea751b5 |
| Aaronson | 2.81 | Not recorded | 68fa77c11e209131cea751b5 |
| Aaronson Cor. | 2.81 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 3.05 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.06 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.2 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 3.29 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 3.33 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.36 | Not recorded | 68fa77c11e209131cea751b5 |
| Kirchenbauer | 3.37 | Not recorded | 68fa77c11e209131cea751b5 |
| Kirchenbauer | 3.39 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 3.41 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.46 | Not recorded | 68fa77c11e209131cea751b5 |
| Random Sampling | 3.5 | Not recorded | 68fa77c11e209131cea751b5 |
| Kuditipudi | 3.55 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.67 | Not recorded | 68fa77c11e209131cea751b5 |
| Kirchenbauer | 3.69 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.72 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.76 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 3.77 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.79 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 3.79 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 3.79 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.8 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 3.86 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 4.06 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 4.06 | Not recorded | 68fa77c11e209131cea751b5 |
| Flat | 4.1 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 4.13 | Not recorded | 68fa77c11e209131cea751b5 |
| Rec. | 4.13 | Not recorded | 68fa77c11e209131cea751b5 |
| Kirchenbauer | 4.67 | Not recorded | 68fa77c11e209131cea751b5 |
| Kirchenbauer | 5.81 | Not recorded | 68fa77c11e209131cea751b5 |

### Greedy Decoding

Perplexity (PPL): 1.37. Result ID: 699e3d755698665ec4fa5c4b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

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

### Flat

Perplexity (PPL): 2.61. Result ID: 699e3d755698665ec4fa5c5a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":1024}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 2.63. Result ID: 699e3d755698665ec4fa5c59.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":512}. This record does not assert independent reproduction.

### Aaronson

Perplexity (PPL): 2.81. Result ID: 699e3d755698665ec4fa5c4d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

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

### Aaronson Cor.

Perplexity (PPL): 2.81. Result ID: 699e3d755698665ec4fa5c4e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

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

### Rec.

Perplexity (PPL): 3.05. Result ID: 699e3d755698665ec4fa5c66.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":512}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.06. Result ID: 699e3d755698665ec4fa5c58.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":32}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.2. Result ID: 699e3d755698665ec4fa5c57.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":16}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 3.29. Result ID: 699e3d755698665ec4fa5c65.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":32}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 3.33. Result ID: 699e3d755698665ec4fa5c64.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":16}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.36. Result ID: 699e3d755698665ec4fa5c56.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":4}. This record does not assert independent reproduction.

### Kirchenbauer

Perplexity (PPL): 3.37. Result ID: 699e3d755698665ec4fa5c51.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73131e209131cea5a4b7.md>)

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

### Kirchenbauer

Perplexity (PPL): 3.39. Result ID: 699e3d755698665ec4fa5c50.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73131e209131cea5a4b7.md>)

Recorded evaluation configuration: {"delta":0.5}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 3.41. Result ID: 699e3d755698665ec4fa5c63.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":4}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.46. Result ID: 699e3d755698665ec4fa5c55.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":1,"m":2}. This record does not assert independent reproduction.

### Random Sampling

Perplexity (PPL): 3.5. Result ID: 699e3d755698665ec4fa5c4c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

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

### Kuditipudi

Perplexity (PPL): 3.55. Result ID: 699e3d755698665ec4fa5c4f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738a1e209131cea5d13e.md>)

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

### Flat

Perplexity (PPL): 3.67. Result ID: 699e3d755698665ec4fa5c62.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":50,"m":32}. This record does not assert independent reproduction.

### Kirchenbauer

Perplexity (PPL): 3.69. Result ID: 699e3d755698665ec4fa5c52.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73131e209131cea5a4b7.md>)

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

### Flat

Perplexity (PPL): 3.72. Result ID: 699e3d755698665ec4fa5c61.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":50,"m":16}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.76. Result ID: 699e3d755698665ec4fa5c60.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":50,"m":4}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 3.77. Result ID: 699e3d755698665ec4fa5c6b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":50,"m":16}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.79. Result ID: 699e3d755698665ec4fa5c5f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":50,"m":2}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 3.79. Result ID: 699e3d755698665ec4fa5c6a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":50,"m":4}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 3.79. Result ID: 699e3d755698665ec4fa5c6c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":50,"m":32}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.8. Result ID: 699e3d755698665ec4fa5c5e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":10,"m":32}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 3.86. Result ID: 699e3d755698665ec4fa5c5d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":10,"m":16}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 4.06. Result ID: 699e3d755698665ec4fa5c5c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":10,"m":4}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 4.06. Result ID: 699e3d755698665ec4fa5c69.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":10,"m":32}. This record does not assert independent reproduction.

### Flat

Perplexity (PPL): 4.1. Result ID: 699e3d755698665ec4fa5c5b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":10,"m":2}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 4.13. Result ID: 699e3d755698665ec4fa5c67.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":10,"m":4}. This record does not assert independent reproduction.

### Rec.

Perplexity (PPL): 4.13. Result ID: 699e3d755698665ec4fa5c68.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

Recorded evaluation configuration: {"k":10,"m":16}. This record does not assert independent reproduction.

### Kirchenbauer

Perplexity (PPL): 4.67. Result ID: 699e3d755698665ec4fa5c53.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73131e209131cea5a4b7.md>)

Recorded evaluation configuration: {"delta":3}. This record does not assert independent reproduction.

### Kirchenbauer

Perplexity (PPL): 5.81. Result ID: 699e3d755698665ec4fa5c54.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77c11e209131cea751b5.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73131e209131cea5a4b7.md>)

Recorded evaluation configuration: {"delta":4}. This record does not assert independent reproduction.

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