resnet_comparison.txt

Resnet18, lr : 0.1, batch size : 8, trained_layers : 10, n_outputs : 32
train :  Epoch 0 ,accuracy : 0.2037, time 7m 27s
train :  Epoch 1 ,accuracy : 0.2707, time 7m 33s
val :  Epoch 0 ,accuracy : 0.2565, time 0m 54s
val :  Epoch 1 ,accuracy : 0.2744, time 0m 54s
Training complete in 16m 49s 
Best val Acc: 0.274386 

Resnet34, lr : 0.1, batch size : 8, trained_layers : 10, n_outputs : 32
train :  Epoch 0 ,accuracy : 0.2066, time 12m 12s
train :  Epoch 1 ,accuracy : 0.2827, time 12m 10s
val :  Epoch 0 ,accuracy : 0.2688, time 0m 53s
val :  Epoch 1 ,accuracy : 0.2796, time 0m 39s
Training complete in 25m 55s 
Best val Acc: 0.279649 

Resnet50, lr : 0.1, batch size : 8, trained_layers : 10, n_outputs : 32
train :  Epoch 0 ,accuracy : 0.2150, time 18m 4s
train :  Epoch 1 ,accuracy : 0.2879, time 17m 57s
val :  Epoch 0 ,accuracy : 0.2709, time 0m 54s
val :  Epoch 1 ,accuracy : 0.2812, time 0m 54s
Training complete in 37m 49s 
Best val Acc: 0.281228 

Resnet101, lr : 0.1, batch size : 8, trained_layers : 10, n_outputs : 32
train :  Epoch 0 ,accuracy : 0.2149, time 32m 6s
train :  Epoch 1 ,accuracy : 0.2981, time 31m 49s
val :  Epoch 0 ,accuracy : 0.2747, time 1m 18s
val :  Epoch 1 ,accuracy : 0.2974, time 1m 17s
Training complete in 66m 31s 
Best val Acc: 0.297368