# 4D Object Question Answering on Objaverse 4,000 object IDs (test) Source: [4D Object Question Answering on Objaverse 4,000 object IDs (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:45:46.528Z Evaluation of 4D Object Question Answering capabilities using the specific test split comprising 4,000 object IDs from the Objaverse dataset. Metrics focus on detailed aspects like spatial relationships, temporal understanding, and appearance. 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 | 4D Object Question Answering | | Dataset (including any stated split) | Objaverse 4,000 object IDs (test) | | Metric | GPT-4 Score | | 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 - [GPT-4 Score](): Higher is better; unit: PERCENT. - [Counting Acc](): Higher is better; unit: PERCENT. - [Temporal Relationship Acc](): Higher is better; unit: PERCENT. - [Action Acc](): Higher is better; unit: PERCENT. - [Spatial Relationship Acc](): Higher is better; unit: PERCENT. - [Appearance Acc](): Higher is better; unit: PERCENT. ## Recorded results | Method | GPT-4 Score | Result date | Report paper | | --- | --- | --- | --- | | 4DPC^2hat | 78.01 | Not recorded | 698423f3a8b01e76cc8ee638 | | MiniGPT-3D | 59.08 | Not recorded | 698423f3a8b01e76cc8ee638 | | ShapeLLM-13B | 56.17 | Not recorded | 698423f3a8b01e76cc8ee638 | | ShapeLLM-7B | 54.58 | Not recorded | 698423f3a8b01e76cc8ee638 | | PointLLM-13B | 54.15 | Not recorded | 698423f3a8b01e76cc8ee638 | | PointLLM-7B | 52.69 | Not recorded | 698423f3a8b01e76cc8ee638 | ### 4DPC^2hat GPT-4 Score: 78.01. Result ID: 6997522e5698665ec4a3fd48. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Input":"3D Point Cloud Sequence"}. This record does not assert independent reproduction. ### MiniGPT-3D GPT-4 Score: 59.08. Result ID: 6997522e5698665ec4a3fd47. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Input":"3D+TA (Temporal Aggregation)"}. This record does not assert independent reproduction. ### ShapeLLM-13B GPT-4 Score: 56.17. Result ID: 6997522e5698665ec4a3fd46. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"13B","Input":"3D+TA (Temporal Aggregation)"}. This record does not assert independent reproduction. ### ShapeLLM-7B GPT-4 Score: 54.58. Result ID: 6997522e5698665ec4a3fd45. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"7B","Input":"3D+TA (Temporal Aggregation)"}. This record does not assert independent reproduction. ### PointLLM-13B GPT-4 Score: 54.15. Result ID: 6997522e5698665ec4a3fd44. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"13B","Input":"3D+TA (Temporal Aggregation)"}. This record does not assert independent reproduction. ### PointLLM-7B GPT-4 Score: 52.69. Result ID: 6997522e5698665ec4a3fd43. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"7B","Input":"3D+TA (Temporal Aggregation)"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.