# LLM Steering on AxBench Source: [LLM Steering on AxBench]() Collection index: [llms.txt]() Data updated: 2026-05-29T10:53:48.869Z Evaluation of LLM Steering capabilities using the AxBench dataset, reporting the steering 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 | LLM Steering | | Dataset (including any stated split) | AxBench | | Metric | Steering Score | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 18 of 18 | | 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 - [Steering Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | Steering Score | Result date | Report paper | | --- | --- | --- | --- | | UNISTEER | 0.74 | Not recorded | 6a191de3dbd7150506f82bf4 | | UNISTEER | 0.68 | Not recorded | 6a191de3dbd7150506f82bf4 | | LoReFT | 0.6 | Not recorded | 6a191de3dbd7150506f82bf4 | | LoReFT | 0.56 | Not recorded | 6a191de3dbd7150506f82bf4 | | LoReFT | 0.42 | Not recorded | 6a191de3dbd7150506f82bf4 | | UNISTEER | 0.36 | Not recorded | 6a191de3dbd7150506f82bf4 | | CAA | 0.14 | Not recorded | 6a191de3dbd7150506f82bf4 | | CAA | 0.11 | Not recorded | 6a191de3dbd7150506f82bf4 | | RepE | 0.11 | Not recorded | 6a191de3dbd7150506f82bf4 | | ODESteer | 0.1 | Not recorded | 6a191de3dbd7150506f82bf4 | | RepE | 0.08 | Not recorded | 6a191de3dbd7150506f82bf4 | | ODESteer | 0.08 | Not recorded | 6a191de3dbd7150506f82bf4 | | ODESteer | 0.08 | Not recorded | 6a191de3dbd7150506f82bf4 | | RepE | 0.05 | Not recorded | 6a191de3dbd7150506f82bf4 | | CAA | 0.04 | Not recorded | 6a191de3dbd7150506f82bf4 | | Original | 0 | Not recorded | 6a191de3dbd7150506f82bf4 | | Original | 0 | Not recorded | 6a191de3dbd7150506f82bf4 | | Original | 0 | Not recorded | 6a191de3dbd7150506f82bf4 | ### UNISTEER Steering Score: 0.74. Result ID: 6a196fdcfbc373592c148c93. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-1.5B"}. This record does not assert independent reproduction. ### UNISTEER Steering Score: 0.68. Result ID: 6a196fdcfbc373592c148c99. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### LoReFT Steering Score: 0.6. Result ID: 6a196fdcfbc373592c148c97. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### LoReFT Steering Score: 0.56. Result ID: 6a196fdcfbc373592c148c91. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-1.5B"}. This record does not assert independent reproduction. ### LoReFT Steering Score: 0.42. Result ID: 6a196fdcfbc373592c148c8b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Llama-3.2-1B"}. This record does not assert independent reproduction. ### UNISTEER Steering Score: 0.36. Result ID: 6a196fdcfbc373592c148c8d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Llama-3.2-1B"}. This record does not assert independent reproduction. ### CAA Steering Score: 0.14. Result ID: 6a196fdcfbc373592c148c8f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-1.5B"}. This record does not assert independent reproduction. ### CAA Steering Score: 0.11. Result ID: 6a196fdcfbc373592c148c95. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### RepE Steering Score: 0.11. Result ID: 6a196fdcfbc373592c148c96. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### ODESteer Steering Score: 0.1. Result ID: 6a196fdcfbc373592c148c92. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-1.5B"}. This record does not assert independent reproduction. ### RepE Steering Score: 0.08. Result ID: 6a196fdcfbc373592c148c8a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Target LLM":"Llama-3.2-1B"}. This record does not assert independent reproduction. ### ODESteer Steering Score: 0.08. Result ID: 6a196fdcfbc373592c148c8c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Llama-3.2-1B"}. This record does not assert independent reproduction. ### ODESteer Steering Score: 0.08. Result ID: 6a196fdcfbc373592c148c98. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-7B"}. This record does not assert independent reproduction. ### RepE Steering Score: 0.05. Result ID: 6a196fdcfbc373592c148c90. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-1.5B"}. This record does not assert independent reproduction. ### CAA Steering Score: 0.04. Result ID: 6a196fdcfbc373592c148c89. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target LLM":"Llama-3.2-1B"}. This record does not assert independent reproduction. ### Original Steering Score: 0. Result ID: 6a196fdcfbc373592c148c88. [Reporting paper]() Recorded evaluation configuration: {"Target LLM":"Llama-3.2-1B"}. This record does not assert independent reproduction. ### Original Steering Score: 0. Result ID: 6a196fdcfbc373592c148c8e. [Reporting paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-1.5B"}. This record does not assert independent reproduction. ### Original Steering Score: 0. Result ID: 6a196fdcfbc373592c148c94. [Reporting paper]() Recorded evaluation configuration: {"Target LLM":"Qwen2.5-7B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.