hand_solver.prototxt

net: "hand_trainval.prototxt"
#test_iter: 100
#test_interval: 1400
base_lr: 0.00033
lr_policy: "step"
gamma: 0.333
stepsize: 500
display: 10
max_iter: 1500
momentum: 0.9
weight_decay: 0.0005
snapshot: 800
snapshot_prefix: "models/hand"