# Preference Modeling on HH-RLHF Source: [Preference Modeling on HH-RLHF]() Collection index: [llms.txt]() Data updated: 2026-05-27T09:20:58.015Z Evaluation of model performance on Preference Modeling using the HH-RLHF dataset, measured by Accuracy. 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 | Preference Modeling | | Dataset (including any stated split) | HH-RLHF | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 30 of 30 | | 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](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | LLM-Judge | 61.4 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | UAB | 60.8 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | UAB | 60.8 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | UAB | 60.6 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | LLM-Judge | 60.2 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Uniform | 60.1 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | N=1 | 60 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Uniform | 59.7 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | N=1 | 59.4 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Uniform | 59.4 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | N=1 | 59.1 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | LLM-Judge | 58.7 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | UAB | 54.3 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | UAB | 53.2 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Length | 53 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | LLM-Judge | 53 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | N=1 | 52.8 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Uniform | 52.2 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | UAB | 51.7 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Length | 51.3 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Random | 51.3 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | N=1 | 50.9 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Random | 50.5 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Uniform | 50.5 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Length | 50.4 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | LLM-Judge | 50.2 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | LLM-Judge | 49.2 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Uniform | 48.2 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | Random | 46.6 | Not recorded | 6a167a8bdbd7150506f7ffc3 | | N=1 | 45.2 | Not recorded | 6a167a8bdbd7150506f7ffc3 | ### LLM-Judge Accuracy: 61.4. Result ID: 6a16b769fbc373592c1024d3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"GPT-OSS-20B","Budget":"N=4"}. This record does not assert independent reproduction. ### UAB Accuracy: 60.8. Result ID: 6a16b769fbc373592c1024d8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Gemma3-27B","Budget":"N=4"}. This record does not assert independent reproduction. ### UAB Accuracy: 60.8. Result ID: 6a16b769fbc373592c1024dc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Cohere","Budget":"N=4"}. This record does not assert independent reproduction. ### UAB Accuracy: 60.6. Result ID: 6a16b769fbc373592c1024d4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"GPT-OSS-20B","Budget":"N=4"}. This record does not assert independent reproduction. ### LLM-Judge Accuracy: 60.2. Result ID: 6a16b769fbc373592c1024d7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Gemma3-27B","Budget":"N=4"}. This record does not assert independent reproduction. ### Uniform Accuracy: 60.1. Result ID: 6a16b769fbc373592c1024d2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"GPT-OSS-20B","Budget":"N=4"}. This record does not assert independent reproduction. ### N=1 Accuracy: 60. Result ID: 6a16b769fbc373592c1024d9. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Cohere","Budget":"N=1"}. This record does not assert independent reproduction. ### Uniform Accuracy: 59.7. Result ID: 6a16b769fbc373592c1024da. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Cohere","Budget":"N=4"}. This record does not assert independent reproduction. ### N=1 Accuracy: 59.4. Result ID: 6a16b769fbc373592c1024d1. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"GPT-OSS-20B","Budget":"N=1"}. This record does not assert independent reproduction. ### Uniform Accuracy: 59.4. Result ID: 6a16b769fbc373592c1024d6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Gemma3-27B","Budget":"N=4"}. This record does not assert independent reproduction. ### N=1 Accuracy: 59.1. Result ID: 6a16b769fbc373592c1024d5. [Reporting paper]() Recorded evaluation configuration: {"Backbone Model":"Gemma3-27B","Budget":"N=1"}. This record does not assert independent reproduction. ### LLM-Judge Accuracy: 58.7. Result ID: 6a16b769fbc373592c1024db. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Cohere","Budget":"N=4"}. This record does not assert independent reproduction. ### UAB Accuracy: 54.3. Result ID: 6a16b769fbc373592c102512. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama3.2-3B","N":"4"}. This record does not assert independent reproduction. ### UAB Accuracy: 53.2. Result ID: 6a16b769fbc373592c102518. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B","N":"4"}. This record does not assert independent reproduction. ### Length Accuracy: 53. Result ID: 6a16b769fbc373592c102515. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B","N":"4"}. This record does not assert independent reproduction. ### LLM-Judge Accuracy: 53. Result ID: 6a16b769fbc373592c102517. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B","N":"4"}. This record does not assert independent reproduction. ### N=1 Accuracy: 52.8. Result ID: 6a16b769fbc373592c102513. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B","N":"1"}. This record does not assert independent reproduction. ### Uniform Accuracy: 52.2. Result ID: 6a16b769fbc373592c102510. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama3.2-3B","N":"4"}. This record does not assert independent reproduction. ### UAB Accuracy: 51.7. Result ID: 6a16b769fbc373592c10250c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B","N":"4"}. This record does not assert independent reproduction. ### Length Accuracy: 51.3. Result ID: 6a16b769fbc373592c10250f. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama3.2-3B","N":"4"}. This record does not assert independent reproduction. ### Random Accuracy: 51.3. Result ID: 6a16b769fbc373592c102514. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B","N":"4"}. This record does not assert independent reproduction. ### N=1 Accuracy: 50.9. Result ID: 6a16b769fbc373592c102507. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B","N":"1"}. This record does not assert independent reproduction. ### Random Accuracy: 50.5. Result ID: 6a16b769fbc373592c10250e. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama3.2-3B","N":"4"}. This record does not assert independent reproduction. ### Uniform Accuracy: 50.5. Result ID: 6a16b769fbc373592c102516. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-7B","N":"4"}. This record does not assert independent reproduction. ### Length Accuracy: 50.4. Result ID: 6a16b769fbc373592c102509. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B","N":"4"}. This record does not assert independent reproduction. ### LLM-Judge Accuracy: 50.2. Result ID: 6a16b769fbc373592c10250b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B","N":"4"}. This record does not assert independent reproduction. ### LLM-Judge Accuracy: 49.2. Result ID: 6a16b769fbc373592c102511. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama3.2-3B","N":"4"}. This record does not assert independent reproduction. ### Uniform Accuracy: 48.2. Result ID: 6a16b769fbc373592c10250a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B","N":"4"}. This record does not assert independent reproduction. ### Random Accuracy: 46.6. Result ID: 6a16b769fbc373592c102508. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-1.5B","N":"4"}. This record does not assert independent reproduction. ### N=1 Accuracy: 45.2. Result ID: 6a16b769fbc373592c10250d. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama3.2-3B","N":"1"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.