# Science Question Answering on ScienceQA (SQAI) Source: [Science Question Answering on ScienceQA (SQAI)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:29:17.674Z Evaluation of model performance on the ScienceQA (SQAI) dataset for the Science Question Answering task, measured using 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) | ScienceQA (SQAI) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 11 of 11 | | 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 | | --- | --- | --- | --- | | LLaVA-1.5 | 71.6 | Not recorded | 68fa715d1e209131cea50df6 | | LVIS-INSTRUCT4V + mixLLaVA (Ours) | 70.6 | Not recorded | 68fa715d1e209131cea50df6 | | LVIS-INSTRUCT4V (Ours) | 69 | Not recorded | 68fa715d1e209131cea50df6 | | LVIS-INSTRUCT4V (Ours) | 68.4 | Not recorded | 68fa715d1e209131cea50df6 | | LVIS-INSTRUCT4V + mixLLaVA (Ours) | 68.3 | Not recorded | 68fa715d1e209131cea50df6 | | Qwen-VL-chat | 68.2 | Not recorded | 68fa715d1e209131cea50df6 | | Qwen-VL | 67.1 | Not recorded | 68fa715d1e209131cea50df6 | | LLaVA-1.5 | 66.8 | Not recorded | 68fa715d1e209131cea50df6 | | InstructBLIP | 63.1 | Not recorded | 68fa715d1e209131cea50df6 | | BLIP-2 | 61 | Not recorded | 68fa715d1e209131cea50df6 | | InstructBLIP | 60.5 | Not recorded | 68fa715d1e209131cea50df6 | ### LLaVA-1.5 Accuracy: 71.6. Result ID: 6991ae203a6c31484909dd3f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-13B","w/G":"GPT-4"}. This record does not assert independent reproduction. ### LVIS-INSTRUCT4V + mixLLaVA (Ours) Accuracy: 70.6. Result ID: 6991ae203a6c31484909dd41. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-13B","w/G":"GPT-4 & 4V"}. This record does not assert independent reproduction. ### LVIS-INSTRUCT4V (Ours) Accuracy: 69. Result ID: 6991ae203a6c31484909dd40. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-13B","w/G":"GPT-4V"}. This record does not assert independent reproduction. ### LVIS-INSTRUCT4V (Ours) Accuracy: 68.4. Result ID: 6991ae203a6c31484909dd3b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-7B","w/G":"GPT-4V"}. This record does not assert independent reproduction. ### LVIS-INSTRUCT4V + mixLLaVA (Ours) Accuracy: 68.3. Result ID: 6991ae203a6c31484909dd3c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-7B","w/G":"GPT-4 & 4V"}. This record does not assert independent reproduction. ### Qwen-VL-chat Accuracy: 68.2. Result ID: 6991ae203a6c31484909dd39. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Qwen-7B","w/G":"Private"}. This record does not assert independent reproduction. ### Qwen-VL Accuracy: 67.1. Result ID: 6991ae203a6c31484909dd38. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Qwen-7B","w/G":"Private"}. This record does not assert independent reproduction. ### LLaVA-1.5 Accuracy: 66.8. Result ID: 6991ae203a6c31484909dd3a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-7B","w/G":"GPT-4"}. This record does not assert independent reproduction. ### InstructBLIP Accuracy: 63.1. Result ID: 6991ae203a6c31484909dd3e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-13B","w/G":"GPT-4"}. This record does not assert independent reproduction. ### BLIP-2 Accuracy: 61. Result ID: 6991ae203a6c31484909dd3d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-13B","w/G":"None"}. This record does not assert independent reproduction. ### InstructBLIP Accuracy: 60.5. Result ID: 6991ae203a6c31484909dd37. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"LLM":"Vicuna-7B","w/G":"GPT-4"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.