| 71 | } |
| 72 | |
| 73 | void test() { |
| 74 | |
| 75 | |
| 76 | |
| 77 | auto lines = load_titles(); |
| 78 | auto gts = load_gt(); |
| 79 | |
| 80 | vector<double> elapsed_times; |
| 81 | for (int loop = 0; loop < 3; loop++) { |
| 82 | FlashBertTokenizer tokenizer("../res/vocab_char_16424.txt", true); |
| 83 | std::chrono::system_clock::time_point t_beg, t_end; |
| 84 | std::chrono::duration<double> diff{}; |
| 85 | |
| 86 | t_beg = std::chrono::system_clock::now(); |
| 87 | |
| 88 | int correct = 0; |
| 89 | long long int total = 0; |
| 90 | for (int i = 0; i < lines.size(); i++) { |
| 91 | auto ids = tokenizer(lines[i]); |
| 92 | if (ids == gts[i]) { |
| 93 | correct += 1; |
| 94 | } |
| 95 | total += ids.size(); |
| 96 | } |
| 97 | t_end = std::chrono::system_clock::now(); |
| 98 | diff = t_end - t_beg; |
| 99 | auto elapsed_time = diff.count(); |
| 100 | elapsed_times.push_back(elapsed_time); |
| 101 | std::cout << elapsed_time << " seconds" << "\t"; |
| 102 | |
| 103 | //std::cout << lines.size() << "\t"; |
| 104 | std::cout << static_cast<double>(correct) * 100.0 / lines.size() << " % Accuracy" << std::endl; |
| 105 | std::cout << static_cast<double>(lines.size()) / elapsed_time << " RPS" << std::endl; |
| 106 | std::cout << "--------------" << std::endl; |
| 107 | } |
| 108 | |
| 109 | double total = std::accumulate(elapsed_times.begin(),elapsed_times.end(),0.0) / 3; |
| 110 | |
| 111 | std::cout << "Final: " << total << std::endl; |
| 112 | } |
| 113 | |
| 114 | int main() { |
| 115 | std::ios::sync_with_stdio(false); |