yonghenglh6
commented
7 years ago Please check the train accuracy of the net in https://github.com/shicai/MobileNet-Caffe. I can only assure there is no calculation error compared with origin caffe.
It seems the lr_policy keeps your learning rate still large when 200000 epoch. Have u ever tried a little max_iter?
Can you share your solver.prototxt ? I only get 43% val acc when the epoch is 200000. Mine is net: "train_val.prototxt" test_initialization: false test_iter: 100 test_interval: 5000 display: 100 average_loss: 40 base_lr: 0.01 lr_policy: "poly" power: 1.0 max_iter: 1000000 momentum: 0.9 weight_decay: 0.0001 snapshot: 5000 snapshot_prefix: "models/mobilenet"