# Science Question Answering on QASC (test) Source: [Science Question Answering on QASC (test)]() Collection index: [llms.txt]() Data updated: 2026-03-06T09:04:58.998Z Evaluation of model performance on the test split of the QASC dataset for Science Question Answering, 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 | Science Question Answering | | Dataset (including any stated split) | QASC (test) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | 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 | | --- | --- | --- | --- | | Imagine-DeBERTa-v3-L | 73.5 | Not recorded | 69aa5f8d2d603e6de3e418cb | | Imagine-DeBERTa-v3-L (Retrieval) | 73.5 | Not recorded | 69aa5f8d2d603e6de3e418cb | | Imagine-DeBERTa-v3-L | 72.4 | Not recorded | 69aa5f8d2d603e6de3e418cb | | CAR-DeBERTa-v3-L | 70 | Not recorded | 69aa5f8d2d603e6de3e418cb | | Imagine-RoBERTa-L | 57.1 | Not recorded | 69aa5f8d2d603e6de3e418cb | | CAR-RoBERTa-L | 56.7 | Not recorded | 69aa5f8d2d603e6de3e418cb | | Imagine-GPT-2-L | 46.5 | Not recorded | 69aa5f8d2d603e6de3e418cb | | Z-LaVI (OPT-30B) | 42.1 | Not recorded | 69aa5f8d2d603e6de3e418cb | | OPT-30B | 39.7 | Not recorded | 69aa5f8d2d603e6de3e418cb | | GPT-J-6B | 36.3 | Not recorded | 69aa5f8d2d603e6de3e418cb | | GPT-Neo-2.7B | 29.6 | Not recorded | 69aa5f8d2d603e6de3e418cb | | Z-LaVI (BART-L) | 27.3 | Not recorded | 69aa5f8d2d603e6de3e418cb | | Z-LaVI (RoBERTa-L) | 27.2 | Not recorded | 69aa5f8d2d603e6de3e418cb | | SMLM | 26.6 | Not recorded | 69aa5f8d2d603e6de3e418cb | ### Imagine-DeBERTa-v3-L Accuracy: 73.5. Result ID: 69aa98b36dbb7075b9548635. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"Synthetic VQA+","setting":"zero-shot"}. This record does not assert independent reproduction. ### Imagine-DeBERTa-v3-L (Retrieval) Accuracy: 73.5. Result ID: 69aa98b36dbb7075b9548636. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"Synthetic VQA+","inference\_mode":"Retrieval","setting":"zero-shot"}. This record does not assert independent reproduction. ### Imagine-DeBERTa-v3-L Accuracy: 72.4. Result ID: 69aa98b36dbb7075b9548634. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"Synthetic VQA","setting":"zero-shot"}. This record does not assert independent reproduction. ### CAR-DeBERTa-v3-L Accuracy: 70. Result ID: 69aa98b36dbb7075b954862d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"AbsAT","setting":"zero-shot"}. This record does not assert independent reproduction. ### Imagine-RoBERTa-L Accuracy: 57.1. Result ID: 69aa98b36dbb7075b9548633. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"Synthetic VQA","setting":"zero-shot"}. This record does not assert independent reproduction. ### CAR-RoBERTa-L Accuracy: 56.7. Result ID: 69aa98b36dbb7075b954862c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"AbsAT","setting":"zero-shot"}. This record does not assert independent reproduction. ### Imagine-GPT-2-L Accuracy: 46.5. Result ID: 69aa98b36dbb7075b9548632. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"Synthetic VQA","setting":"zero-shot"}. This record does not assert independent reproduction. ### Z-LaVI (OPT-30B) Accuracy: 42.1. Result ID: 69aa98b36dbb7075b9548631. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"setting":"zero-shot"}. This record does not assert independent reproduction. ### OPT-30B Accuracy: 39.7. Result ID: 69aa98b36dbb7075b954862e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"setting":"zero-shot"}. This record does not assert independent reproduction. ### GPT-J-6B Accuracy: 36.3. Result ID: 69aa98b36dbb7075b954862b. [Reporting paper]() Recorded evaluation configuration: {"setting":"zero-shot"}. This record does not assert independent reproduction. ### GPT-Neo-2.7B Accuracy: 29.6. Result ID: 69aa98b36dbb7075b954862a. [Reporting paper]() Recorded evaluation configuration: {"setting":"zero-shot"}. This record does not assert independent reproduction. ### Z-LaVI (BART-L) Accuracy: 27.3. Result ID: 69aa98b36dbb7075b9548630. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"setting":"zero-shot"}. This record does not assert independent reproduction. ### Z-LaVI (RoBERTa-L) Accuracy: 27.2. Result ID: 69aa98b36dbb7075b954862f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"setting":"zero-shot"}. This record does not assert independent reproduction. ### SMLM Accuracy: 26.6. Result ID: 69aa98b36dbb7075b9548629. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"KB":"\*","setting":"zero-shot"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.