step down results, set up for exp and sigmoid decay

.gitattributes

@@ -0,0 +1,2 @@
+# Auto detect text files and perform LF normalization
+* text=auto

cars/lr-investigations/exponential/1e-2/0.95/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/exponential/1e-2/0.95/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/exponential/1e-2/0.95/solver.prototxt

@@ -0,0 +1,14 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.00999999977648
+display: 12
+max_iter: 10200
+lr_policy: "exp"
+gamma: 0.999497234821
+momentum: 0.899999976158
+weight_decay: 9.99999974738e-05
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/exponential/1e-2/0.95/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/exponential/1e-2/0.98/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/exponential/1e-2/0.98/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/exponential/1e-2/0.98/solver.prototxt

@@ -0,0 +1,14 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.00999999977648
+display: 12
+max_iter: 10200
+lr_policy: "exp"
+gamma: 0.999801933765
+momentum: 0.899999976158
+weight_decay: 9.99999974738e-05
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/exponential/1e-2/0.98/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/exponential/1e-2/0.99/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/exponential/1e-2/0.99/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/exponential/1e-2/0.99/solver.prototxt

@@ -0,0 +1,14 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.00999999977648
+display: 12
+max_iter: 10200
+lr_policy: "exp"
+gamma: 0.999901473522
+momentum: 0.899999976158
+weight_decay: 9.99999974738e-05
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/exponential/1e-2/0.99/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-MAN-2/digits/jobs/20210407-214532-d396/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.99999974738e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.00499999988824
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.10000000149
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.00999999977648
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/sigmoid/1e-2/50_0.05/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/sigmoid/1e-2/50_0.05/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/sigmoid/1e-2/50_0.05/solver.prototxt

@@ -0,0 +1,15 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.0099999998
+display: 12
+max_iter: 10200
+lr_policy: "sigmoid"
+gamma: -0.00049019611
+momentum: 0.89999998
+weight_decay: 9.9999997e-05
+stepsize: 5100
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/sigmoid/1e-2/50_0.05/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/sigmoid/1e-2/50_0.1/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/sigmoid/1e-2/50_0.1/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/sigmoid/1e-2/50_0.1/solver.prototxt

@@ -0,0 +1,15 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.0099999998
+display: 12
+max_iter: 10200
+lr_policy: "sigmoid"
+gamma: -0.00098039221
+momentum: 0.89999998
+weight_decay: 9.9999997e-05
+stepsize: 5100
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/sigmoid/1e-2/50_0.1/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/sigmoid/1e-2/50_0.2/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/sigmoid/1e-2/50_0.2/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/sigmoid/1e-2/50_0.2/solver.prototxt

@@ -0,0 +1,15 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.0099999998
+display: 12
+max_iter: 10200
+lr_policy: "sigmoid"
+gamma: -0.0019607844
+momentum: 0.89999998
+weight_decay: 9.9999997e-05
+stepsize: 5100
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/sigmoid/1e-2/50_0.2/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-TAM-1/digits/jobs/20210407-221849-d51c/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/step-down/1e-2/33_0.1/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/step-down/1e-2/33_0.1/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/step-down/1e-2/33_0.1/solver.prototxt

@@ -0,0 +1,15 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.0099999998
+display: 12
+max_iter: 10200
+lr_policy: "step"
+gamma: 0.1
+momentum: 0.89999998
+weight_decay: 9.9999997e-05
+stepsize: 3366
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/step-down/1e-2/33_0.1/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/step-down/1e-2/33_0.25/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/step-down/1e-2/33_0.25/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/step-down/1e-2/33_0.25/solver.prototxt

@@ -0,0 +1,15 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.0099999998
+display: 12
+max_iter: 10200
+lr_policy: "step"
+gamma: 0.25
+momentum: 0.89999998
+weight_decay: 9.9999997e-05
+stepsize: 3366
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/step-down/1e-2/33_0.25/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/step-down/1e-2/33_0.5/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/step-down/1e-2/33_0.5/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/step-down/1e-2/33_0.5/solver.prototxt

@@ -0,0 +1,15 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.0099999998
+display: 12
+max_iter: 10200
+lr_policy: "step"
+gamma: 0.5
+momentum: 0.89999998
+weight_decay: 9.9999997e-05
+stepsize: 3366
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/step-down/1e-2/33_0.5/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

cars/lr-investigations/step-down/1e-2/33_0.75/deploy.prototxt

@@ -0,0 +1,341 @@
+input: "data"
+input_shape {
+  dim: 1
+  dim: 3
+  dim: 227
+  dim: 227
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+}

cars/lr-investigations/step-down/1e-2/33_0.75/original.prototxt

@@ -0,0 +1,388 @@
+name: "AlexNet"
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "train"
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 128
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    stage: "val"
+  }
+  transform_param {
+    crop_size: 227
+  }
+  data_param {
+    batch_size: 32
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    stage: "val"
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+  exclude {
+    stage: "deploy"
+  }
+}
+layer {
+  name: "softmax"
+  type: "Softmax"
+  bottom: "fc8"
+  top: "softmax"
+  include {
+    stage: "deploy"
+  }
+}

cars/lr-investigations/step-down/1e-2/33_0.75/solver.prototxt

@@ -0,0 +1,15 @@
+test_iter: 51
+test_interval: 102
+base_lr: 0.0099999998
+display: 12
+max_iter: 10200
+lr_policy: "step"
+gamma: 0.75
+momentum: 0.89999998
+weight_decay: 9.9999997e-05
+stepsize: 3366
+snapshot: 102
+snapshot_prefix: "snapshot"
+solver_mode: GPU
+net: "train_val.prototxt"
+solver_type: SGD

cars/lr-investigations/step-down/1e-2/33_0.75/train_val.prototxt

@@ -0,0 +1,382 @@
+layer {
+  name: "train-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TRAIN
+  }
+  transform_param {
+    mirror: true
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/train_db"
+    batch_size: 128
+    backend: LMDB
+  }
+}
+layer {
+  name: "val-data"
+  type: "Data"
+  top: "data"
+  top: "label"
+  include {
+    phase: TEST
+  }
+  transform_param {
+    crop_size: 227
+    mean_file: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/mean.binaryproto"
+  }
+  data_param {
+    source: "/mnt/bigdisk/DIGITS-AIN-3/digits/jobs/20210401-115716-aaf7/val_db"
+    batch_size: 32
+    backend: LMDB
+  }
+}
+layer {
+  name: "conv1"
+  type: "Convolution"
+  bottom: "data"
+  top: "conv1"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 96
+    kernel_size: 11
+    stride: 4
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu1"
+  type: "ReLU"
+  bottom: "conv1"
+  top: "conv1"
+}
+layer {
+  name: "norm1"
+  type: "LRN"
+  bottom: "conv1"
+  top: "norm1"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool1"
+  type: "Pooling"
+  bottom: "norm1"
+  top: "pool1"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv2"
+  type: "Convolution"
+  bottom: "pool1"
+  top: "conv2"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 2
+    kernel_size: 5
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu2"
+  type: "ReLU"
+  bottom: "conv2"
+  top: "conv2"
+}
+layer {
+  name: "norm2"
+  type: "LRN"
+  bottom: "conv2"
+  top: "norm2"
+  lrn_param {
+    local_size: 5
+    alpha: 9.9999997e-05
+    beta: 0.75
+  }
+}
+layer {
+  name: "pool2"
+  type: "Pooling"
+  bottom: "norm2"
+  top: "pool2"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "conv3"
+  type: "Convolution"
+  bottom: "pool2"
+  top: "conv3"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "relu3"
+  type: "ReLU"
+  bottom: "conv3"
+  top: "conv3"
+}
+layer {
+  name: "conv4"
+  type: "Convolution"
+  bottom: "conv3"
+  top: "conv4"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 384
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu4"
+  type: "ReLU"
+  bottom: "conv4"
+  top: "conv4"
+}
+layer {
+  name: "conv5"
+  type: "Convolution"
+  bottom: "conv4"
+  top: "conv5"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  convolution_param {
+    num_output: 256
+    pad: 1
+    kernel_size: 3
+    group: 2
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu5"
+  type: "ReLU"
+  bottom: "conv5"
+  top: "conv5"
+}
+layer {
+  name: "pool5"
+  type: "Pooling"
+  bottom: "conv5"
+  top: "pool5"
+  pooling_param {
+    pool: MAX
+    kernel_size: 3
+    stride: 2
+  }
+}
+layer {
+  name: "fc6"
+  type: "InnerProduct"
+  bottom: "pool5"
+  top: "fc6"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu6"
+  type: "ReLU"
+  bottom: "fc6"
+  top: "fc6"
+}
+layer {
+  name: "drop6"
+  type: "Dropout"
+  bottom: "fc6"
+  top: "fc6"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc7"
+  type: "InnerProduct"
+  bottom: "fc6"
+  top: "fc7"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 4096
+    weight_filler {
+      type: "gaussian"
+      std: 0.0049999999
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.1
+    }
+  }
+}
+layer {
+  name: "relu7"
+  type: "ReLU"
+  bottom: "fc7"
+  top: "fc7"
+}
+layer {
+  name: "drop7"
+  type: "Dropout"
+  bottom: "fc7"
+  top: "fc7"
+  dropout_param {
+    dropout_ratio: 0.5
+  }
+}
+layer {
+  name: "fc8"
+  type: "InnerProduct"
+  bottom: "fc7"
+  top: "fc8"
+  param {
+    lr_mult: 1.0
+    decay_mult: 1.0
+  }
+  param {
+    lr_mult: 2.0
+    decay_mult: 0.0
+  }
+  inner_product_param {
+    num_output: 196
+    weight_filler {
+      type: "gaussian"
+      std: 0.0099999998
+    }
+    bias_filler {
+      type: "constant"
+      value: 0.0
+    }
+  }
+}
+layer {
+  name: "accuracy"
+  type: "Accuracy"
+  bottom: "fc8"
+  bottom: "label"
+  top: "accuracy"
+  include {
+    phase: TEST
+  }
+}
+layer {
+  name: "loss"
+  type: "SoftmaxWithLoss"
+  bottom: "fc8"
+  bottom: "label"
+  top: "loss"
+}

pyproject.toml

@@ -5,7 +5,7 @@ description = ""
 authors = ["andy <andy@sarsoo.xyz>"]
 
 [tool.poetry.dependencies]
-python = "^3.9"
+python = "^3.8"
 numpy = "^1.20.2"
 matplotlib = "^3.4.1"
 jupyterlab = "^3.0.12"