# Question Answering on Administrative Dataset (test) Source: [Question Answering on Administrative Dataset (test)]() Collection index: [llms.txt]() Data updated: 2026-05-15T07:32:27.481Z Evaluation of Question Answering performance on the test split of the Administrative Dataset, reporting metrics such as acceptance rate and average rank statistics. 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 | Question Answering | | Dataset (including any stated split) | Administrative Dataset (test) | | Metric | Acceptance Rate | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 6 of 6 | | 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 - [Acceptance Rate](): Higher is better; unit: PERCENT. - [Rank 1 Score](): Higher is better; unit: SCALAR. - [Rank 2 Score](): Higher is better; unit: SCALAR. - [Rank 3 Score](): Higher is better; unit: SCALAR. - [Rank 4 Score](): Higher is better; unit: SCALAR. - [Rank 5 Score](): Higher is better; unit: SCALAR. - [Average Rank](): Lower is better; unit: SCALAR. - [Std Dev of Rank](): Lower is better; unit: SCALAR. ## Recorded results | Method | Acceptance Rate | Result date | Report paper | | --- | --- | --- | --- | | Derivation Prompting | 89.6 | Not recorded | 6a06a899dbc836b87d82a054 | | Derivation Prompting | 82.2 | Not recorded | 6a06a899dbc836b87d82a054 | | RAG | 77.8 | Not recorded | 6a06a899dbc836b87d82a054 | | Long context | 76.3 | Not recorded | 6a06a899dbc836b87d82a054 | | RAG | 72.6 | Not recorded | 6a06a899dbc836b87d82a054 | | Long context | 65.2 | Not recorded | 6a06a899dbc836b87d82a054 | ### Derivation Prompting Acceptance Rate: 89.6. Result ID: 6a06cc0347f34ce924ee0cd4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Claude Opus","Evaluator":"Claude Opus","Context (k)":"3","Retrieval":"Cross-Encoder","Prompting Strategy":"Derivation Prompting"}. This record does not assert independent reproduction. ### Derivation Prompting Acceptance Rate: 82.2. Result ID: 6a06cc0347f34ce924ee0cd1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Claude Haiku","Evaluator":"Claude Opus","Context (k)":"3","Retrieval":"Cross-Encoder","Prompting Strategy":"Derivation Prompting"}. This record does not assert independent reproduction. ### RAG Acceptance Rate: 77.8. Result ID: 6a06cc0347f34ce924ee0cd3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Claude Opus","Evaluator":"Claude Opus","Context (k)":"3","Retrieval":"Cross-Encoder","Prompting Strategy":"RAG"}. This record does not assert independent reproduction. ### Long context Acceptance Rate: 76.3. Result ID: 6a06cc0347f34ce924ee0cd2. [Reporting paper]() Recorded evaluation configuration: {"Model":"Claude Opus","Evaluator":"Claude Opus","Context (k)":"3","Prompting Strategy":"Long context"}. This record does not assert independent reproduction. ### RAG Acceptance Rate: 72.6. Result ID: 6a06cc0347f34ce924ee0cd0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Claude Haiku","Evaluator":"Claude Opus","Context (k)":"3","Retrieval":"Cross-Encoder","Prompting Strategy":"RAG"}. This record does not assert independent reproduction. ### Long context Acceptance Rate: 65.2. Result ID: 6a06cc0347f34ce924ee0ccf. [Reporting paper]() Recorded evaluation configuration: {"Model":"Claude Haiku","Evaluator":"Claude Opus","Context (k)":"3","Prompting Strategy":"Long context"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.