```json { "table_name": "benchmarks", "table_type_tags": [ {"tag": "dimension", "confidence": 0.95} ], "primary_type": "dimension", "insufficient_evidence": false, "entity": { "columns": ["id_runs"], "unique_per_row": true }, "time_axis": null, "reasoning": "id_runs has Uniqueness 1.000 (1311 unique values in 1311 rows), making each row a unique benchmark run. The table catalogs evaluation results: each row describes one model's performance on one task at one point in time. Timestamp columns (started_at, Version release date, Publication date) are attributes of the run/model, not an axis ordering state changes. The entity is unique per row with no repeating axis → dimension." } ```