inception_solver.prototxt

net: "models/googlenet/inception_v3_train_val.prototxt"
test_iter: 1000
test_interval: 4000
test_initialization: false
display: 40
average_loss: 40
base_lr: 0.01
lr_policy: "poly"
stepsize: 320000
gamma: 0.96
max_iter: 1000000
power: 1.0
momentum: 0.9
weight_decay: 0.0002
snapshot: 40000
snapshot_prefix: "models/googlenet/inception_v3_"
solver_mode: GPU