| 68 | } |
| 69 | |
| 70 | virtual void InitTinyNet(const bool force_backward = false, |
| 71 | const bool accuracy_layer = false) { |
| 72 | string proto = |
| 73 | "name: 'TinyTestNetwork' " |
| 74 | "layer { " |
| 75 | " name: 'data' " |
| 76 | " type: 'DummyData' " |
| 77 | " dummy_data_param { " |
| 78 | " shape { " |
| 79 | " dim: 5 " |
| 80 | " dim: 2 " |
| 81 | " dim: 3 " |
| 82 | " dim: 4 " |
| 83 | " } " |
| 84 | " data_filler { " |
| 85 | " type: 'gaussian' " |
| 86 | " std: 0.01 " |
| 87 | " } " |
| 88 | " shape { " |
| 89 | " dim: 5 " |
| 90 | " } " |
| 91 | " data_filler { " |
| 92 | " type: 'constant' " |
| 93 | " value: 0 " |
| 94 | " } " |
| 95 | " } " |
| 96 | " top: 'data' " |
| 97 | " top: 'label' " |
| 98 | "} " |
| 99 | "layer { " |
| 100 | " name: 'innerproduct' " |
| 101 | " type: 'InnerProduct' " |
| 102 | " inner_product_param { " |
| 103 | " num_output: 1000 " |
| 104 | " weight_filler { " |
| 105 | " type: 'gaussian' " |
| 106 | " std: 0.01 " |
| 107 | " } " |
| 108 | " bias_filler { " |
| 109 | " type: 'constant' " |
| 110 | " value: 0 " |
| 111 | " } " |
| 112 | " } " |
| 113 | " param { " |
| 114 | " lr_mult: 1 " |
| 115 | " decay_mult: 1 " |
| 116 | " } " |
| 117 | " param { " |
| 118 | " lr_mult: 2 " |
| 119 | " decay_mult: 0 " |
| 120 | " } " |
| 121 | " bottom: 'data' " |
| 122 | " top: 'innerproduct' " |
| 123 | "} " |
| 124 | "layer { " |
| 125 | " name: 'loss' " |
| 126 | " type: 'SoftmaxWithLoss' " |
| 127 | " bottom: 'innerproduct' " |