pcam-interpretability/dino-vits16-val08842-vit-v2

dino-vits16

Best Validation Accuracy: 0.8842

Metadata

• Model Name: dino-vits16
• Optimizer: adamw
• Scheduler: cosine
• Weight Decay: 0.02
• Warmup Epochs: 15
• Patience: 10
• Amp: True
• Seed: 42
• Batch Size: 352
• Initial Lr: 0.001
• Total Epochs Ran: 15
• Early Stopped: True
• Training Time Seconds: 8274.155663728714
• Num Parameters: 21666049
• Device: NVIDIA A100-SXM4-40GB
• Run Id: vit-v2

Training Configuration

• Epochs: 15
• Batch size: 352
• Learning rate (initial): 0.001

Training Logs (Per Epoch)

Epoch	Train Loss	Train Acc	Val Loss	Val Acc	LR
1	0.4232	0.8257	0.3356	0.8533	0.000067
2	0.2757	0.8853	0.2970	0.8798	0.000133
3	0.2358	0.9042	0.3283	0.8632	0.000200
4	0.2285	0.9069	0.3900	0.8392	0.000267
5	0.2238	0.9095	0.2741	0.8842	0.000333
6	0.2216	0.9105	0.3205	0.8721	0.000400
7	0.2163	0.9131	0.3025	0.8762	0.000467
8	0.2157	0.9133	0.2963	0.8738	0.000533
9	0.2124	0.9145	0.3113	0.8667	0.000600
10	0.2111	0.9152	0.3100	0.8674	0.000667
11	0.2094	0.9159	0.3786	0.8381	0.000733
12	0.2059	0.9174	0.3727	0.8452	0.000800
13	0.2056	0.9172	0.2960	0.8794	0.000867
14	0.2000	0.9198	0.3909	0.8299	0.000933
15	0.1993	0.9201	0.3634	0.8526	0.001000