| 70 | |
| 71 | template <typename Dtype> |
| 72 | void LSTMUnitLayer<Dtype>::Backward_cpu(const vector<Blob<Dtype>*>& top, |
| 73 | const vector<bool>& propagate_down, const vector<Blob<Dtype>*>& bottom) { |
| 74 | CHECK(!propagate_down[2]) << "Cannot backpropagate to sequence indicators."; |
| 75 | if (!propagate_down[0] && !propagate_down[1]) { return; } |
| 76 | |
| 77 | const int num = bottom[0]->shape(1); |
| 78 | const int x_dim = hidden_dim_ * 4; |
| 79 | const Dtype* C_prev = bottom[0]->cpu_data(); |
| 80 | const Dtype* X = bottom[1]->cpu_data(); |
| 81 | const Dtype* cont = bottom[2]->cpu_data(); |
| 82 | const Dtype* C = top[0]->cpu_data(); |
| 83 | const Dtype* H = top[1]->cpu_data(); |
| 84 | const Dtype* C_diff = top[0]->cpu_diff(); |
| 85 | const Dtype* H_diff = top[1]->cpu_diff(); |
| 86 | Dtype* C_prev_diff = bottom[0]->mutable_cpu_diff(); |
| 87 | Dtype* X_diff = bottom[1]->mutable_cpu_diff(); |
| 88 | for (int n = 0; n < num; ++n) { |
| 89 | for (int d = 0; d < hidden_dim_; ++d) { |
| 90 | const Dtype i = sigmoid(X[d]); |
| 91 | const Dtype f = (*cont == 0) ? 0 : |
| 92 | (*cont * sigmoid(X[1 * hidden_dim_ + d])); |
| 93 | const Dtype o = sigmoid(X[2 * hidden_dim_ + d]); |
| 94 | const Dtype g = tanh(X[3 * hidden_dim_ + d]); |
| 95 | const Dtype c_prev = C_prev[d]; |
| 96 | const Dtype c = C[d]; |
| 97 | const Dtype tanh_c = tanh(c); |
| 98 | Dtype* c_prev_diff = C_prev_diff + d; |
| 99 | Dtype* i_diff = X_diff + d; |
| 100 | Dtype* f_diff = X_diff + 1 * hidden_dim_ + d; |
| 101 | Dtype* o_diff = X_diff + 2 * hidden_dim_ + d; |
| 102 | Dtype* g_diff = X_diff + 3 * hidden_dim_ + d; |
| 103 | const Dtype c_term_diff = |
| 104 | C_diff[d] + H_diff[d] * o * (1 - tanh_c * tanh_c); |
| 105 | *c_prev_diff = c_term_diff * f; |
| 106 | *i_diff = c_term_diff * g * i * (1 - i); |
| 107 | *f_diff = c_term_diff * c_prev * f * (1 - f); |
| 108 | *o_diff = H_diff[d] * tanh_c * o * (1 - o); |
| 109 | *g_diff = c_term_diff * i * (1 - g * g); |
| 110 | } |
| 111 | C_prev += hidden_dim_; |
| 112 | X += x_dim; |
| 113 | C += hidden_dim_; |
| 114 | H += hidden_dim_; |
| 115 | C_diff += hidden_dim_; |
| 116 | H_diff += hidden_dim_; |
| 117 | X_diff += x_dim; |
| 118 | C_prev_diff += hidden_dim_; |
| 119 | ++cont; |
| 120 | } |
| 121 | } |
| 122 | |
| 123 | #ifdef CPU_ONLY |
| 124 | STUB_GPU(LSTMUnitLayer); |