happy8825/siglip_classification_result_20251217_011529

SigLIP Classification Eval Model: happy8825/siglip-ecva-main Test repo: happy8825/siglip_test (split: train) Samples visualized: 50 of 924 Metrics accuracy: 0.5790 f1: 0.5482 total: 924, correct: 535 Evaluation Runtime dataset size: 924 total time: 24.64 sec avg infer time (per sample): 6.97 ms GPU memory avg: 363.5 MiB GPU memory peak: 370.9 MiB Confusion Matrix