# LLM Judgement Confidence Estimation on AlpacaEval (test) Source: [LLM Judgement Confidence Estimation on AlpacaEval (test)]() Collection index: [llms.txt]() Data updated: 2026-05-18T07:07:51.778Z Evaluation of LLM confidence estimation capabilities using the test split of the AlpacaEval dataset, measured by Rank Correlation (RK) and 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) | AlpacaEval (test) | | Metric | Rank Correlation (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 - [Rank Correlation (RK)](): Higher is better; unit: SCALAR. - [AUROC](): Higher is better; unit: PERCENT. ## Recorded results | Method | Rank Correlation (RK) | Result date | Report paper | | --- | --- | --- | --- | | Verbalized Confidence | 0.4367 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Random Annotator | 0.4321 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Predictive Probability | 0.4214 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Simulated Annotators | 0.4177 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Predictive Probability | 0.4025 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Predictive Probability | 0.4022 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Simulated Annotators | 0.3922 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Random Annotator | 0.392 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Random Annotator | 0.3867 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Learning Confidence (Vanilla) | 0.3865 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Simulated Annotators | 0.3839 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Margin-Adaptive Confidence Ranking | 0.3393 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Learning Confidence (Vanilla) | 0.337 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Learning Confidence (Vanilla) | 0.3236 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Margin-Adaptive Confidence Ranking | 0.2776 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | | Margin-Adaptive Confidence Ranking | 0.2707 | Not recorded | 6a0a9d11dbc836b87d82b3e2 | ### Verbalized Confidence Rank Correlation (RK): 0.4367. Result ID: 6a0abaae47f34ce924f05dde. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Random Annotator Rank Correlation (RK): 0.4321. Result ID: 6a0abaae47f34ce924f05ddf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Predictive Probability Rank Correlation (RK): 0.4214. Result ID: 6a0abaae47f34ce924f05ddd. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Simulated Annotators Rank Correlation (RK): 0.4177. Result ID: 6a0abaae47f34ce924f05de0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Predictive Probability Rank Correlation (RK): 0.4025. Result ID: 6a0abaae47f34ce924f05de8. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Predictive Probability Rank Correlation (RK): 0.4022. Result ID: 6a0abaae47f34ce924f05de3. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Simulated Annotators Rank Correlation (RK): 0.3922. Result ID: 6a0abaae47f34ce924f05dea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Random Annotator Rank Correlation (RK): 0.392. Result ID: 6a0abaae47f34ce924f05de9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Random Annotator Rank Correlation (RK): 0.3867. Result ID: 6a0abaae47f34ce924f05de4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Learning Confidence (Vanilla) Rank Correlation (RK): 0.3865. Result ID: 6a0abaae47f34ce924f05de1. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Simulated Annotators Rank Correlation (RK): 0.3839. Result ID: 6a0abaae47f34ce924f05de5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Margin-Adaptive Confidence Ranking Rank Correlation (RK): 0.3393. Result ID: 6a0abaae47f34ce924f05de2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Mistral-7B"}. This record does not assert independent reproduction. ### Learning Confidence (Vanilla) Rank Correlation (RK): 0.337. Result ID: 6a0abaae47f34ce924f05deb. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Qwen2.5-72B"}. This record does not assert independent reproduction. ### Learning Confidence (Vanilla) Rank Correlation (RK): 0.3236. Result ID: 6a0abaae47f34ce924f05de6. [Reporting paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Margin-Adaptive Confidence Ranking Rank Correlation (RK): 0.2776. Result ID: 6a0abaae47f34ce924f05de7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Judge Model":"Llama3-70B"}. This record does not assert independent reproduction. ### Margin-Adaptive Confidence Ranking Rank Correlation (RK): 0.2707. Result ID: 6a0abaae47f34ce924f05dec. [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.