=== ViT-Mini frozen baselines === --- ViT-Mini seed=42 --- ViT-Mini frozen: 16266/809354 trainable params [ViT-frozen] s=42 ep 10: acc=0.5014 [ViT-frozen] s=42 ep 20: acc=0.5297 [ViT-frozen] s=42 ep 30: acc=0.5375 [ViT-frozen] s=42 ep 40: acc=0.5554 [ViT-frozen] s=42 ep 50: acc=0.5619 [ViT-frozen] s=42 ep 60: acc=0.5678 ViT-Mini shallow: 16266 params (no blocks) [ViT-shallow] s=42 ep 10: acc=0.1000 [ViT-shallow] s=42 ep 20: acc=0.1000 [ViT-shallow] s=42 ep 30: acc=0.1000 [ViT-shallow] s=42 ep 40: acc=0.1000 [ViT-shallow] s=42 ep 50: acc=0.1000 [ViT-shallow] s=42 ep 60: acc=0.1000 FINAL ViT s=42: frozen=0.5678, shallow=0.1000 --- ViT-Mini seed=123 --- ViT-Mini frozen: 16266/809354 trainable params [ViT-frozen] s=123 ep 10: acc=0.5249 [ViT-frozen] s=123 ep 20: acc=0.5421 [ViT-frozen] s=123 ep 30: acc=0.5545 [ViT-frozen] s=123 ep 40: acc=0.5621 [ViT-frozen] s=123 ep 50: acc=0.5680 [ViT-frozen] s=123 ep 60: acc=0.5728 ViT-Mini shallow: 16266 params (no blocks) [ViT-shallow] s=123 ep 10: acc=0.1000 [ViT-shallow] s=123 ep 20: acc=0.1000 [ViT-shallow] s=123 ep 30: acc=0.1000 [ViT-shallow] s=123 ep 40: acc=0.1000 [ViT-shallow] s=123 ep 50: acc=0.1000 [ViT-shallow] s=123 ep 60: acc=0.1000 FINAL ViT s=123: frozen=0.5728, shallow=0.1000 --- ViT-Mini seed=456 --- ViT-Mini frozen: 16266/809354 trainable params [ViT-frozen] s=456 ep 10: acc=0.5166 [ViT-frozen] s=456 ep 20: acc=0.5342 [ViT-frozen] s=456 ep 30: acc=0.5414 [ViT-frozen] s=456 ep 40: acc=0.5545 [ViT-frozen] s=456 ep 50: acc=0.5615 [ViT-frozen] s=456 ep 60: acc=0.5681 ViT-Mini shallow: 16266 params (no blocks) [ViT-shallow] s=456 ep 10: acc=0.1000 [ViT-shallow] s=456 ep 20: acc=0.1000 [ViT-shallow] s=456 ep 30: acc=0.1000 [ViT-shallow] s=456 ep 40: acc=0.1000 [ViT-shallow] s=456 ep 50: acc=0.1000 [ViT-shallow] s=456 ep 60: acc=0.1000 FINAL ViT s=456: frozen=0.5681, shallow=0.1000 === StudentNet frozen baselines === --- StudentNet seed=42 --- StudentNet frozen: 1290/67850 trainable params [Student-frozen] s=42 ep 10: acc=0.8405 [Student-frozen] s=42 ep 20: acc=0.8860 [Student-frozen] s=42 ep 30: acc=0.8990 [Student-frozen] s=42 ep 40: acc=0.8990 [Student-frozen] s=42 ep 50: acc=0.9015 [Student-frozen] s=42 ep 60: acc=0.9005 [Student-frozen] s=42 ep 70: acc=0.9005 [Student-frozen] s=42 ep 80: acc=0.9005 StudentNet shallow: 1290 params (no blocks) [Student-shallow] s=42 ep 10: acc=0.8555 [Student-shallow] s=42 ep 20: acc=0.8990 [Student-shallow] s=42 ep 30: acc=0.9085 [Student-shallow] s=42 ep 40: acc=0.9140 [Student-shallow] s=42 ep 50: acc=0.9140 [Student-shallow] s=42 ep 60: acc=0.9155 [Student-shallow] s=42 ep 70: acc=0.9165 [Student-shallow] s=42 ep 80: acc=0.9165 FINAL Student s=42: frozen=0.9005, shallow=0.9165 --- StudentNet seed=123 --- StudentNet frozen: 1290/67850 trainable params [Student-frozen] s=123 ep 10: acc=0.8505 [Student-frozen] s=123 ep 20: acc=0.8915 [Student-frozen] s=123 ep 30: acc=0.9030 [Student-frozen] s=123 ep 40: acc=0.9050 [Student-frozen] s=123 ep 50: acc=0.9050 [Student-frozen] s=123 ep 60: acc=0.9060 [Student-frozen] s=123 ep 70: acc=0.9060 [Student-frozen] s=123 ep 80: acc=0.9060 StudentNet shallow: 1290 params (no blocks) [Student-shallow] s=123 ep 10: acc=0.8495 [Student-shallow] s=123 ep 20: acc=0.8900 [Student-shallow] s=123 ep 30: acc=0.9050 [Student-shallow] s=123 ep 40: acc=0.9080 [Student-shallow] s=123 ep 50: acc=0.9105 [Student-shallow] s=123 ep 60: acc=0.9125 [Student-shallow] s=123 ep 70: acc=0.9130 [Student-shallow] s=123 ep 80: acc=0.9135 FINAL Student s=123: frozen=0.9060, shallow=0.9135 --- StudentNet seed=456 --- StudentNet frozen: 1290/67850 trainable params [Student-frozen] s=456 ep 10: acc=0.8300 [Student-frozen] s=456 ep 20: acc=0.8900 [Student-frozen] s=456 ep 30: acc=0.9025 [Student-frozen] s=456 ep 40: acc=0.9115 [Student-frozen] s=456 ep 50: acc=0.9150 [Student-frozen] s=456 ep 60: acc=0.9165 [Student-frozen] s=456 ep 70: acc=0.9180 [Student-frozen] s=456 ep 80: acc=0.9175 StudentNet shallow: 1290 params (no blocks) [Student-shallow] s=456 ep 10: acc=0.8615 [Student-shallow] s=456 ep 20: acc=0.9050 [Student-shallow] s=456 ep 30: acc=0.9160 [Student-shallow] s=456 ep 40: acc=0.9200 [Student-shallow] s=456 ep 50: acc=0.9235 [Student-shallow] s=456 ep 60: acc=0.9225 [Student-shallow] s=456 ep 70: acc=0.9225 [Student-shallow] s=456 ep 80: acc=0.9225 FINAL Student s=456: frozen=0.9175, shallow=0.9225 Saved: results/frozen_baselines_crossarch.json