# Instruction Adherence and Security Robustness on StruQ Clean 1.0

Source: [Instruction Adherence and Security Robustness on StruQ Clean 1.0](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0>)

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

Data updated: 2026-02-26T05:38:56.216Z

Evaluation of model performance on the StruQ Clean 1.0 dataset, focusing on instruction adherence and security robustness, measured by various Attack Success Rate (ASR) variants and a general capability 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 | Instruction Adherence and Security Robustness |
| Dataset (including any stated split) | StruQ Clean 1.0 |
| Metric | Capability Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| 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

- [Capability Score](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0.md?metric=Capability%20Score>): Higher is better; unit: PERCENT.

- [ASR (Naive)](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0.md?metric=ASR%20(Naive)>): Lower is better; unit: PERCENT.

- [ASR (Ignore)](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0.md?metric=ASR%20(Ignore)>): Lower is better; unit: PERCENT.

- [ASR (Escape-S)](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0.md?metric=ASR%20(Escape-S)>): Lower is better; unit: PERCENT.

- [ASR (Completion-R)](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0.md?metric=ASR%20(Completion-R)>): Lower is better; unit: PERCENT.

- [ASR (Average)](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0.md?metric=ASR%20(Average)>): Lower is better; unit: PERCENT.

- [ASR (Worst)](<https://www.sota2.com/research/sota/instruction-adherence-and-security-robustness-on-struq-clean-1-0.md?metric=ASR%20(Worst)>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Capability Score | Result date | Report paper |
| --- | --- | --- | --- |
| Text | 84.87 | Not recorded | 69311580d6b0dd3342933b0d |
| Delimiter | 84.23 | Not recorded | 69311580d6b0dd3342933b0d |
| CAHL | 83.6 | Not recorded | 69311580d6b0dd3342933b0d |
| ISE | 78.64 | Not recorded | 69311580d6b0dd3342933b0d |

### Text

Capability Score: 84.87. Result ID: 699708e35698665ec4563a0a.

[Reporting paper](<https://www.sota2.com/research/paper/69311580d6b0dd3342933b0d.md>)

Recorded evaluation configuration: {"Training Type":"Vanilla delimiter"}. This record does not assert independent reproduction.

### Delimiter

Capability Score: 84.23. Result ID: 699708e35698665ec4563a0b.

[Reporting paper](<https://www.sota2.com/research/paper/69311580d6b0dd3342933b0d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74fb1e209131cea65117.md>)

Recorded evaluation configuration: {"Training Type":"Special delimiter from StruQ"}. This record does not assert independent reproduction.

### CAHL

Capability Score: 83.6. Result ID: 699708e35698665ec4563a0d.

[Reporting paper](<https://www.sota2.com/research/paper/69311580d6b0dd3342933b0d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69311580d6b0dd3342933b0d.md>)

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

### ISE

Capability Score: 78.64. Result ID: 699708e35698665ec4563a0c.

[Reporting paper](<https://www.sota2.com/research/paper/69311580d6b0dd3342933b0d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78db1e209131cea7ad67.md>)

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

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