# Privacy-preserving Query Rewriting on DelegateCI-Bench General

Source: [Privacy-preserving Query Rewriting on DelegateCI-Bench General](<https://www.sota2.com/research/sota/privacy-preserving-query-rewriting-on-delegateci-bench-general>)

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

Data updated: 2026-06-04T10:54:05.204Z

Evaluation of privacy-preserving query rewriting performance on the DelegateCI-Bench General dataset, using metrics such as SLR, NE, EA, and Utilization.

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 | Privacy-preserving Query Rewriting |
| Dataset (including any stated split) | DelegateCI-Bench General |
| Metric | Utilization |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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

- [Utilization](<https://www.sota2.com/research/sota/privacy-preserving-query-rewriting-on-delegateci-bench-general.md?metric=Utilization>): Higher is better; unit: PERCENT.

- [SLR](<https://www.sota2.com/research/sota/privacy-preserving-query-rewriting-on-delegateci-bench-general.md?metric=SLR>): Higher is better; unit: SCALAR.

- [NE](<https://www.sota2.com/research/sota/privacy-preserving-query-rewriting-on-delegateci-bench-general.md?metric=NE>): Lower is better; unit: SCALAR.

- [EA](<https://www.sota2.com/research/sota/privacy-preserving-query-rewriting-on-delegateci-bench-general.md?metric=EA>): Higher is better; unit: PERCENT.

- [Average Utilization](<https://www.sota2.com/research/sota/privacy-preserving-query-rewriting-on-delegateci-bench-general.md?metric=Average%20Utilization>): Higher is better; unit: PERCENT.

- [Delta vs Local](<https://www.sota2.com/research/sota/privacy-preserving-query-rewriting-on-delegateci-bench-general.md?metric=Delta%20vs%20Local>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Utilization | Result date | Report paper |
| --- | --- | --- | --- |
| Direct LLM | 100 | Not recorded | 6a2106dec10484bac23f0508 |
| PUFT | 96 | Not recorded | 6a2106dec10484bac23f0508 |
| CI-guided reinforcement learning framework | 95.3 | Not recorded | 6a2106dec10484bac23f0508 |
| CI-guided reinforcement learning framework | 94.3 | Not recorded | 6a2106dec10484bac23f0508 |
| Papillon | 93.5 | Not recorded | 6a2106dec10484bac23f0508 |
| PUFT | 92.7 | Not recorded | 6a2106dec10484bac23f0508 |
| Papillon | 92.2 | Not recorded | 6a2106dec10484bac23f0508 |
| Local Only | 89.9 | Not recorded | 6a2106dec10484bac23f0508 |
| CI-guided reinforcement learning framework | 89.5 | Not recorded | 6a2106dec10484bac23f0508 |
| Presidio(PII Only) | 89.2 | Not recorded | 6a2106dec10484bac23f0508 |
| Local Only | 87.2 | Not recorded | 6a2106dec10484bac23f0508 |
| Papillon | 80.3 | Not recorded | 6a2106dec10484bac23f0508 |
| PUFT | 80 | Not recorded | 6a2106dec10484bac23f0508 |
| Local Only | 71.8 | Not recorded | 6a2106dec10484bac23f0508 |

### Direct LLM

Utilization: 100. Result ID: 6a21593cfbc373592c1de0a8.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"raw query"}. This record does not assert independent reproduction.

### PUFT

Utilization: 96. Result ID: 6a21593cfbc373592c1de0b0.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Qwen-2.5-7B","Local aggregator":"Qwen-2.5-7B-Instruct"}. This record does not assert independent reproduction.

### CI-guided reinforcement learning framework

Utilization: 95.3. Result ID: 6a21593cfbc373592c1de0b1.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Qwen-2.5-3B (shared)","Local aggregator":"Qwen-2.5-7B-Instruct"}. This record does not assert independent reproduction.

### CI-guided reinforcement learning framework

Utilization: 94.3. Result ID: 6a21593cfbc373592c1de0b5.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Qwen-2.5-3B (shared)","Local aggregator":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction.

### Papillon

Utilization: 93.5. Result ID: 6a21593cfbc373592c1de0af.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Qwen-2.5-7B","Local aggregator":"Qwen-2.5-7B-Instruct"}. This record does not assert independent reproduction.

### PUFT

Utilization: 92.7. Result ID: 6a21593cfbc373592c1de0b4.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Llama-3.1-8B","Local aggregator":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction.

### Papillon

Utilization: 92.2. Result ID: 6a21593cfbc373592c1de0b3.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Llama-3.1-8B","Local aggregator":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction.

### Local Only

Utilization: 89.9. Result ID: 6a21593cfbc373592c1de0ae.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"original query","Local aggregator":"Qwen-2.5-7B-Instruct"}. This record does not assert independent reproduction.

### CI-guided reinforcement learning framework

Utilization: 89.5. Result ID: 6a21593cfbc373592c1de0ad.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Qwen-2.5-3B (shared)","Local aggregator":"Qwen-2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

### Presidio(PII Only)

Utilization: 89.2. Result ID: 6a21593cfbc373592c1de0a9.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"rule-based PII"}. This record does not assert independent reproduction.

### Local Only

Utilization: 87.2. Result ID: 6a21593cfbc373592c1de0b2.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"original query","Local aggregator":"Llama-3.1-8B-Instruct"}. This record does not assert independent reproduction.

### Papillon

Utilization: 80.3. Result ID: 6a21593cfbc373592c1de0ab.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Qwen-2.5-1.5B","Local aggregator":"Qwen-2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

### PUFT

Utilization: 80. Result ID: 6a21593cfbc373592c1de0ac.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"Qwen-2.5-1.5B","Local aggregator":"Qwen-2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

### Local Only

Utilization: 71.8. Result ID: 6a21593cfbc373592c1de0aa.

[Reporting paper](<https://www.sota2.com/research/paper/6a2106dec10484bac23f0508.md>)

Recorded evaluation configuration: {"Rewriter":"original query","Local aggregator":"Qwen-2.5-1.5B-Instruct"}. This record does not assert independent reproduction.

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