# Logic Inversion on OpenAgent Tier-3 Source: [Logic Inversion on OpenAgent Tier-3]() Collection index: [llms.txt]() Data updated: 2026-07-02T09:37:28.178Z Evaluation of model performance on the OpenAgent Tier-3 dataset focusing on the Logic Inversion task, measured by Accuracy Delta (Acc Δ). 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 | Logic Inversion | | Dataset (including any stated split) | OpenAgent Tier-3 | | Metric | Accuracy Delta | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 15 of 15 | | 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 - [Accuracy Delta](): Higher is better; unit: SCALAR. ## Recorded results | Method | Accuracy Delta | Result date | Report paper | | --- | --- | --- | --- | | SFT-200 + PAFT | 22.7 | Not recorded | 6a45f097dc71372d7795a3c6 | | SFT-400 + PAFT | -2.8 | Not recorded | 6a45f097dc71372d7795a3c6 | | Base | -8.5 | Not recorded | 6a45f097dc71372d7795a3c6 | | SFT-800 + PAFT | -9.8 | Not recorded | 6a45f097dc71372d7795a3c6 | | PAFT (alpha=0.3 default) | -9.8 | Not recorded | 6a45f097dc71372d7795a3c6 | | PAFT (SRP only) | -10.6 | Not recorded | 6a45f097dc71372d7795a3c6 | | SFT-600 + PAFT | -10.7 | Not recorded | 6a45f097dc71372d7795a3c6 | | PAFT (alpha=0.4) | -11.1 | Not recorded | 6a45f097dc71372d7795a3c6 | | PAFT (alpha=0.2) | -15.8 | Not recorded | 6a45f097dc71372d7795a3c6 | | PAFT (EFP only) | -27 | Not recorded | 6a45f097dc71372d7795a3c6 | | SFT-800 | -28 | Not recorded | 6a45f097dc71372d7795a3c6 | | PAFT (SBP only) | -31.7 | Not recorded | 6a45f097dc71372d7795a3c6 | | SFT-400 | -32.5 | Not recorded | 6a45f097dc71372d7795a3c6 | | SFT-600 | -33 | Not recorded | 6a45f097dc71372d7795a3c6 | | SFT-200 | -39.9 | Not recorded | 6a45f097dc71372d7795a3c6 | ### SFT-200 + PAFT Accuracy Delta: 22.7. Result ID: 6a46315460b88bb339863678. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"200","PAFT":"true"}. This record does not assert independent reproduction. ### SFT-400 + PAFT Accuracy Delta: -2.8. Result ID: 6a46315460b88bb33986367a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"400","PAFT":"true"}. This record does not assert independent reproduction. ### Base Accuracy Delta: -8.5. Result ID: 6a46315460b88bb339863676. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SFT-800 + PAFT Accuracy Delta: -9.8. Result ID: 6a46315460b88bb33986367e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800","PAFT":"true"}. This record does not assert independent reproduction. ### PAFT (alpha=0.3 default) Accuracy Delta: -9.8. Result ID: 6a46315460b88bb339863680. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800","perturbation\_ratio":"0.3"}. This record does not assert independent reproduction. ### PAFT (SRP only) Accuracy Delta: -10.6. Result ID: 6a46315460b88bb339863684. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800","perturbation\_ratio":"0.3","component":"SRP"}. This record does not assert independent reproduction. ### SFT-600 + PAFT Accuracy Delta: -10.7. Result ID: 6a46315460b88bb33986367c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"600","PAFT":"true"}. This record does not assert independent reproduction. ### PAFT (alpha=0.4) Accuracy Delta: -11.1. Result ID: 6a46315460b88bb339863681. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800","perturbation\_ratio":"0.4"}. This record does not assert independent reproduction. ### PAFT (alpha=0.2) Accuracy Delta: -15.8. Result ID: 6a46315460b88bb33986367f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800","perturbation\_ratio":"0.2"}. This record does not assert independent reproduction. ### PAFT (EFP only) Accuracy Delta: -27. Result ID: 6a46315460b88bb339863682. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800","perturbation\_ratio":"0.3","component":"EFP"}. This record does not assert independent reproduction. ### SFT-800 Accuracy Delta: -28. Result ID: 6a46315460b88bb33986367d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800"}. This record does not assert independent reproduction. ### PAFT (SBP only) Accuracy Delta: -31.7. Result ID: 6a46315460b88bb339863683. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"800","perturbation\_ratio":"0.3","component":"SBP"}. This record does not assert independent reproduction. ### SFT-400 Accuracy Delta: -32.5. Result ID: 6a46315460b88bb339863679. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"400"}. This record does not assert independent reproduction. ### SFT-600 Accuracy Delta: -33. Result ID: 6a46315460b88bb33986367b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"600"}. This record does not assert independent reproduction. ### SFT-200 Accuracy Delta: -39.9. Result ID: 6a46315460b88bb339863677. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_steps":"200"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.