# Model Card Generation on MetaGAI-Bench Model Cards Source: [Model Card Generation on MetaGAI-Bench Model Cards]() Collection index: [llms.txt]() Data updated: 2026-04-14T07:20:35.659Z Evaluation of model performance on generating comprehensive and high-quality Model Cards using the MetaGAI-Bench Model Cards dataset, assessed across dimensions like Faithfulness, Relevance, Accuracy, Consistency, and Usefulness. 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 | Model Card Generation | | Dataset (including any stated split) | MetaGAI-Bench Model Cards | | Metric | Faithfulness (G1) Rank | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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 - [Faithfulness (G1) Rank](): Higher is better; unit: SCALAR. - [Faithfulness (G1) Score](): Higher is better; unit: SCALAR. - [Faithfulness (G2) Rank](): Higher is better; unit: SCALAR. - [Faithfulness (G2) Score](): Higher is better; unit: SCALAR. - [Relevance (G1) Rank](): Higher is better; unit: SCALAR. - [Relevance (G1) Score](): Higher is better; unit: SCALAR. - [Relevance (G2) Rank](): Higher is better; unit: SCALAR. - [Relevance (G2) Score](): Higher is better; unit: SCALAR. - [Accuracy (G1) Rank](): Higher is better; unit: SCALAR. - [Accuracy (G1) Score](): Higher is better; unit: SCALAR. - [Accuracy (G2) Rank](): Higher is better; unit: SCALAR. - [Accuracy (G2) Score](): Higher is better; unit: SCALAR. - [Consistency (G1) Rank](): Higher is better; unit: SCALAR. - [Consistency (G1) Score](): Higher is better; unit: SCALAR. - [Consistency (G2) Rank](): Higher is better; unit: SCALAR. - [Consistency (G2) Score](): Higher is better; unit: SCALAR. - [Usefulness (G1) Rank](): Higher is better; unit: SCALAR. - [Usefulness (G1) Score](): Higher is better; unit: SCALAR. - [Usefulness (G2) Rank](): Higher is better; unit: SCALAR. - [Usefulness (G2) Score](): Higher is better; unit: SCALAR. - [Average (G1) Rank](): Higher is better; unit: SCALAR. - [Average (G1) Score](): Higher is better; unit: SCALAR. - [Average (G2) Rank](): Higher is better; unit: SCALAR. - [Average (G2) Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | Faithfulness (G1) Rank | Result date | Report paper | | --- | --- | --- | --- | | Zero-shot | 4 | Not recorded | 69ddca007d0172140ec798eb | | Zero-shot | 4 | Not recorded | 69ddca007d0172140ec798eb | | Human-Annotated | 4 | Not recorded | 69ddca007d0172140ec798eb | | CardGen | 3 | Not recorded | 69ddca007d0172140ec798eb | | Zero-shot | 3 | Not recorded | 69ddca007d0172140ec798eb | | CardGen | 3 | Not recorded | 69ddca007d0172140ec798eb | | AdaQE-CG | 2 | Not recorded | 69ddca007d0172140ec798eb | | CardGen | 2 | Not recorded | 69ddca007d0172140ec798eb | | AdaQE-CG | 2 | Not recorded | 69ddca007d0172140ec798eb | | Human-Annotated | 1 | Not recorded | 69ddca007d0172140ec798eb | | Human-Annotated | 1 | Not recorded | 69ddca007d0172140ec798eb | | AdaQE-CG | 1 | Not recorded | 69ddca007d0172140ec798eb | ### Zero-shot Faithfulness (G1) Rank: 4. Result ID: 69ddea3647f34ce924c4e9be. [Reporting paper]() Recorded evaluation configuration: {"Card Type":"Model"}. This record does not assert independent reproduction. ### Zero-shot Faithfulness (G1) Rank: 4. Result ID: 69ddea3647f34ce924c4e9c6. [Reporting paper]() Recorded evaluation configuration: {"LLM Judge":"GPT-5-nano"}. This record does not assert independent reproduction. ### Human-Annotated Faithfulness (G1) Rank: 4. Result ID: 69ddea3647f34ce924c4e9cb. [Reporting paper]() Recorded evaluation configuration: {"LLM Judge":"Gemini-2.5 Flash-Lite"}. This record does not assert independent reproduction. ### CardGen Faithfulness (G1) Rank: 3. Result ID: 69ddea3647f34ce924c4e9bf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Card Type":"Model"}. This record does not assert independent reproduction. ### Zero-shot Faithfulness (G1) Rank: 3. Result ID: 69ddea3647f34ce924c4e9c7. [Reporting paper]() Recorded evaluation configuration: {"LLM Judge":"Gemini-2.5 Flash-Lite"}. This record does not assert independent reproduction. ### CardGen Faithfulness (G1) Rank: 3. Result ID: 69ddea3647f34ce924c4e9c8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Judge":"GPT-5-nano"}. This record does not assert independent reproduction. ### AdaQE-CG Faithfulness (G1) Rank: 2. Result ID: 69ddea3647f34ce924c4e9c1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Card Type":"Model"}. This record does not assert independent reproduction. ### CardGen Faithfulness (G1) Rank: 2. Result ID: 69ddea3647f34ce924c4e9c9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Judge":"Gemini-2.5 Flash-Lite"}. This record does not assert independent reproduction. ### AdaQE-CG Faithfulness (G1) Rank: 2. Result ID: 69ddea3647f34ce924c4e9cc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Judge":"GPT-5-nano"}. This record does not assert independent reproduction. ### Human-Annotated Faithfulness (G1) Rank: 1. Result ID: 69ddea3647f34ce924c4e9c0. [Reporting paper]() Recorded evaluation configuration: {"Card Type":"Model"}. This record does not assert independent reproduction. ### Human-Annotated Faithfulness (G1) Rank: 1. Result ID: 69ddea3647f34ce924c4e9ca. [Reporting paper]() Recorded evaluation configuration: {"LLM Judge":"GPT-5-nano"}. This record does not assert independent reproduction. ### AdaQE-CG Faithfulness (G1) Rank: 1. Result ID: 69ddea3647f34ce924c4e9cd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Judge":"Gemini-2.5 Flash-Lite"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.