# LLM Judgement Confidence Estimation on HH-RLHF (test) Source: [LLM Judgement Confidence Estimation on HH-RLHF (test)]() Collection index: [llms.txt]() Data updated: 2026-05-18T07:07:51.813Z Evaluation of LLM confidence estimation capabilities using the test split of the HH-RLHF dataset, measured by Rank Correlation (RK) and Area Under the ROC Curve (AUROC). 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 Judgement Confidence Estimation | | Dataset (including any stated split) | HH-RLHF (test) | | Metric | RK | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 - [RK](): Higher is better; unit: SCALAR. - [AUROC](): Higher is better; unit: SCALAR. ## Recorded results | Method | RK | Result date | Report paper | | --- | --- | --- | --- | | Predictive Probability | 0.4763 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Verbalized Confidence | 0.4738 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Predictive Probability | 0.4478 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Predictive Probability | 0.4409 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Simulated Annotators | 0.3986 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Learning Confidence (Vanilla) | 0.3841 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Random Annotator | 0.3803 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Random Annotator | 0.376 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Random Annotator | 0.3751 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Simulated Annotators | 0.3678 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Learning Confidence (Vanilla) | 0.3597 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Simulated Annotators | 0.3571 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Learning Confidence (Vanilla) | 0.3482 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Margin-Adaptive Confidence Ranking | 0.3286 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Margin-Adaptive Confidence Ranking | 0.3094 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Margin-Adaptive Confidence Ranking | 0.2813 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | ### Predictive Probability RK: 0.4763. Result ID: 6a0abaae47f34ce924f05ded. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Verbalized Confidence RK: 0.4738. Result ID: 6a0abaae47f34ce924f05dee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Predictive Probability RK: 0.4478. Result ID: 6a0abaae47f34ce924f05df3. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Predictive Probability RK: 0.4409. Result ID: 6a0abaae47f34ce924f05df8. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Simulated Annotators RK: 0.3986. Result ID: 6a0abaae47f34ce924f05df0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Learning Confidence (Vanilla) RK: 0.3841. Result ID: 6a0abaae47f34ce924f05df1. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Random Annotator RK: 0.3803. Result ID: 6a0abaae47f34ce924f05def. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Random Annotator RK: 0.376. Result ID: 6a0abaae47f34ce924f05df9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Random Annotator RK: 0.3751. Result ID: 6a0abaae47f34ce924f05df4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Simulated Annotators RK: 0.3678. Result ID: 6a0abaae47f34ce924f05dfa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Learning Confidence (Vanilla) RK: 0.3597. Result ID: 6a0abaae47f34ce924f05df6. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Simulated Annotators RK: 0.3571. Result ID: 6a0abaae47f34ce924f05df5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Learning Confidence (Vanilla) RK: 0.3482. Result ID: 6a0abaae47f34ce924f05dfb. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Margin-Adaptive Confidence Ranking RK: 0.3286. Result ID: 6a0abaae47f34ce924f05df2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Margin-Adaptive Confidence Ranking RK: 0.3094. Result ID: 6a0abaae47f34ce924f05df7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Margin-Adaptive Confidence Ranking RK: 0.2813. Result ID: 6a0abaae47f34ce924f05dfc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.