|
| 1 | +#include "bert.h" |
| 2 | +#include "ggml.h" |
| 3 | + |
| 4 | +#include <unistd.h> |
| 5 | +#include <map> |
| 6 | +#include <algorithm> |
| 7 | +#include <stdio.h> |
| 8 | +#include <string> |
| 9 | +#include <vector> |
| 10 | +#include <fstream> |
| 11 | +#include <sstream> |
| 12 | +#define ANSI_COLOR_RED "\x1b[31m" |
| 13 | +#define ANSI_COLOR_RESET "\x1b[0m" |
| 14 | +#define ANSI_COLOR_GREEN "\x1b[32m" |
| 15 | + |
| 16 | + |
| 17 | +std::vector<std::string> txt2list(const std::string& filename) { |
| 18 | + std::ifstream file(filename); |
| 19 | + std::vector<std::string> all_lines; |
| 20 | + |
| 21 | + if (!file.is_open()) { |
| 22 | + printf("can not open file: %s\n", filename.c_str()); |
| 23 | + return all_lines; |
| 24 | + } |
| 25 | + |
| 26 | + std::string line; |
| 27 | + while (std::getline(file, line)) { |
| 28 | + all_lines.push_back(line); |
| 29 | + } |
| 30 | + |
| 31 | + file.close(); |
| 32 | + return all_lines; |
| 33 | +} |
| 34 | + |
| 35 | +std::vector<std::vector<int>> read_expected_tokenids(const std::string& filename) { |
| 36 | + std::ifstream file(filename); |
| 37 | + std::vector<std::vector<int>> all_numbers; |
| 38 | + |
| 39 | + if (!file.is_open()) { |
| 40 | + printf("can not open file: %s\n", filename.c_str()); |
| 41 | + return all_numbers; |
| 42 | + } |
| 43 | + |
| 44 | + |
| 45 | + std::string line; |
| 46 | + while (std::getline(file, line)) { |
| 47 | + std::vector<int> line_numbers; |
| 48 | + std::istringstream iss(line); |
| 49 | + std::string number_str; |
| 50 | + |
| 51 | + while (std::getline(iss, number_str, ',')) { |
| 52 | + line_numbers.push_back(std::stoi(number_str)); |
| 53 | + } |
| 54 | + |
| 55 | + all_numbers.push_back(line_numbers); |
| 56 | + } |
| 57 | + |
| 58 | + file.close(); |
| 59 | + return all_numbers; |
| 60 | +} |
| 61 | + |
| 62 | +void tokenizer_test(bert_ctx * ctx, const std::string& input, const bert_tokens& expected) { |
| 63 | + int N = bert_n_max_tokens(ctx); |
| 64 | + bert_tokens result = bert_tokenize(ctx, input, N); |
| 65 | + int n_tokens; |
| 66 | + |
| 67 | + if (result != expected) { |
| 68 | + printf("tokenizer test failed: '%.*s'\n", 16000, input.data()); |
| 69 | + |
| 70 | + printf("["); |
| 71 | + for (auto& tok : result) { |
| 72 | + printf("%d, ", tok); |
| 73 | + } |
| 74 | + printf("]\n"); |
| 75 | + |
| 76 | + for (size_t i = 0; i < result.size(); i++) { |
| 77 | + bert_token a = expected[std::min(i, expected.size()-1)]; |
| 78 | + bert_token b = result[i]; |
| 79 | + const char *color_start = (a == b) ? ANSI_COLOR_GREEN : ANSI_COLOR_RED; |
| 80 | + const char *color_end = ANSI_COLOR_RESET; |
| 81 | + |
| 82 | + printf("%s%d -> %s : %d -> %s%s\n", color_start, a, bert_vocab_id_to_token(ctx, a), b, bert_vocab_id_to_token(ctx, b), color_end); |
| 83 | + } |
| 84 | + } else { |
| 85 | + printf("Success '%.*s...'\n", 16, input.data()); |
| 86 | + } |
| 87 | +} |
| 88 | + |
| 89 | + |
| 90 | +struct bert_params |
| 91 | +{ |
| 92 | + int32_t n_threads = 6; |
| 93 | + const char* model = "models/all-MiniLM-L6-v2/ggml-model-q4_0.bin"; |
| 94 | + const char* prompt = "test prompt"; |
| 95 | + int32_t batch_size = 32; |
| 96 | + bool use_cpu = false; |
| 97 | +}; |
| 98 | + |
| 99 | +void bert_print_usage(char **argv, const bert_params ¶ms) { |
| 100 | + fprintf(stderr, "usage: %s [options]\n", argv[0]); |
| 101 | + fprintf(stderr, "\n"); |
| 102 | + fprintf(stderr, "options:\n"); |
| 103 | + fprintf(stderr, " -h, --help show this help message and exit\n"); |
| 104 | + fprintf(stderr, " -m FNAME, --model FNAME\n"); |
| 105 | + fprintf(stderr, " model path (default: %s)\n", params.model); |
| 106 | + fprintf(stderr, " batch size to use when executing model\n"); |
| 107 | + fprintf(stderr, " -c, --cpu use CPU backend (default: use CUDA if available)\n"); |
| 108 | + fprintf(stderr, "\n"); |
| 109 | +} |
| 110 | + |
| 111 | +bool bert_params_parse(int argc, char **argv, bert_params ¶ms) { |
| 112 | + for (int i = 1; i < argc; i++) |
| 113 | + { |
| 114 | + std::string arg = argv[i]; |
| 115 | + |
| 116 | + if (arg == "-m" || arg == "--model") { |
| 117 | + params.model = argv[++i]; |
| 118 | + } else if (arg == "-c" || arg == "--cpu") { |
| 119 | + params.use_cpu = true; |
| 120 | + } else if (arg == "-h" || arg == "--help") { |
| 121 | + bert_print_usage(argv, params); |
| 122 | + exit(0); |
| 123 | + } else { |
| 124 | + fprintf(stderr, "error: unknown argument: %s\n", arg.c_str()); |
| 125 | + bert_print_usage(argv, params); |
| 126 | + exit(0); |
| 127 | + } |
| 128 | + } |
| 129 | + |
| 130 | + return true; |
| 131 | +} |
| 132 | + |
| 133 | +int main(int argc, char ** argv) { |
| 134 | + |
| 135 | + bert_params params; |
| 136 | + params.model = "models/all-MiniLM-L6-v2/ggml-model-q4_0.bin"; |
| 137 | + |
| 138 | + if (bert_params_parse(argc, argv, params) == false) { |
| 139 | + return 1; |
| 140 | + } |
| 141 | + |
| 142 | + |
| 143 | + bert_ctx * bctx; |
| 144 | + |
| 145 | + // load the model |
| 146 | + { |
| 147 | + if ((bctx = bert_load_from_file(params.model, params.use_cpu)) == nullptr) { |
| 148 | + fprintf(stderr, "%s: failed to load model from '%s'\n", __func__, params.model); |
| 149 | + return 1; |
| 150 | + } |
| 151 | + } |
| 152 | + std::string dir = params.model; |
| 153 | + std::size_t i = dir.rfind("/models/"); |
| 154 | + if (i != std::string::npos) { |
| 155 | + dir.resize(i); |
| 156 | + } else { |
| 157 | + dir = "."; |
| 158 | + } |
| 159 | + |
| 160 | + auto expected = read_expected_tokenids(dir + "/tests/hf_tokenized_ids.txt"); |
| 161 | + auto prompts = txt2list(dir + "/tests/test_prompts.txt"); |
| 162 | + |
| 163 | + if (expected.size() == 0 || prompts.size() == 0) { |
| 164 | + printf("failed to read test data\n"); |
| 165 | + return 1; |
| 166 | + } |
| 167 | + |
| 168 | + if (expected.size() != prompts.size()) { |
| 169 | + printf("test data size mismatch\n"); |
| 170 | + return 1; |
| 171 | + } |
| 172 | + |
| 173 | + // tokenizer tests: |
| 174 | + for (size_t i = 0; i < prompts.size(); i++) { |
| 175 | + tokenizer_test(bctx, prompts[i], expected[i]); |
| 176 | + } |
| 177 | + |
| 178 | + // tokenizer_test(bctx, "1231 2431431", {101, 13138, 2487, 22884, 16932, 21486, 102}); |
| 179 | + // tokenizer_test(bctx, "Québec", {101, 5447, 102}); |
| 180 | + // tokenizer_test(bctx, "syömme \t täällä tänään", {101, 25353, 5358, 4168, 11937, 25425, 9092, 14634, 102}); |
| 181 | + // tokenizer_test(bctx, "I'm going to the store to buy 3 apples and a banana! You're welcome to come along if you'd like. The time is 2:30 p.m. and it's partly cloudy outside. I'll be back soon, so don't go anywhere.", {101, 1045, 1005, 1049, 2183, 2000, 1996, 3573, 2000, 4965, 1017, 18108, 1998, 1037, 15212, 999, 2017, 1005, 2128, 6160, 2000, 2272, 2247, 2065, 2017, 1005, 1040, 2066, 1012, 1996, 2051, 2003, 1016, 1024, 2382, 1052, 1012, 1049, 1012, 1998, 2009, 1005, 1055, 6576, 24706, 2648, 1012, 1045, 1005, 2222, 2022, 2067, 2574, 1010, 2061, 2123, 1005, 1056, 2175, 5973, 1012, 102}); |
| 182 | + // tokenizer_test(bctx, "\"5 2 + 3 * 4 -\"; int stack[1000], top = -1; int calculate(int a, int b, char operator) { return operator == '+' ? a + b : operator == '-' ? a - b : operator == '*' ? a * b : a / b; } void push(int x) { stack[++top] = x; } int pop() { return stack[top--]; } int evaluatePostfix(char* expression) { for (int i = 0; expression[i]; i++) { if (isdigit(expression[i])) push(expression[i] - '0'); else { int a = pop(), b = pop(); push(calculate(b, a, expression[i])); } } return pop(); } int result = evaluatePostfix(input);", {101, 1000, 1019, 1016, 1009, 1017, 1008, 1018, 1011, 1000, 1025, 20014, 9991, 1031, 6694, 1033, 1010, 2327, 1027, 1011, 1015, 1025, 20014, 18422, 1006, 20014, 1037, 1010, 20014, 1038, 1010, 25869, 6872, 1007, 1063, 2709, 6872, 1027, 1027, 1005, 1009, 1005, 1029, 1037, 1009, 1038, 1024, 6872, 1027, 1027, 1005, 1011, 1005, 1029, 1037, 1011, 1038, 1024, 6872, 1027, 1027, 1005, 1008, 1005, 1029, 1037, 1008, 1038, 1024, 1037, 1013, 1038, 1025, 1065, 11675, 5245, 1006, 20014, 1060, 1007, 1063, 9991, 1031, 1009, 1009, 2327, 1033, 1027, 1060, 1025, 1065, 20014, 3769, 1006, 1007, 1063, 2709, 9991, 1031, 2327, 1011, 1011, 1033, 1025, 1065, 20014, 16157, 19894, 8873, 2595, 1006, 25869, 1008, 3670, 1007, 1063, 2005, 1006, 20014, 1045, 1027, 1014, 1025, 3670, 1031, 1045, 1033, 1025, 1045, 1009, 1009, 1007, 1063, 2065, 1006, 2003, 4305, 23806, 1006, 3670, 1031, 1045, 1033, 1007, 1007, 5245, 1006, 3670, 1031, 1045, 1033, 1011, 1005, 1014, 1005, 1007, 1025, 2842, 1063, 20014, 1037, 1027, 3769, 1006, 1007, 1010, 1038, 1027, 3769, 1006, 1007, 1025, 5245, 1006, 18422, 1006, 1038, 1010, 1037, 1010, 3670, 1031, 1045, 1033, 1007, 1007, 1025, 1065, 1065, 2709, 3769, 1006, 1007, 1025, 1065, 20014, 2765, 1027, 16157, 19894, 8873, 2595, 1006, 7953, 1007, 1025, 102}); |
| 183 | + |
| 184 | + // tokenizer_test(bctx, "Hello world!", {101, 7592, 2088, 999, 102}); |
| 185 | + // tokenizer_test(bctx, "你好，世界！", {101, 100, 100, 1989, 1745, 100, 1986, 102}); |
| 186 | + // tokenizer_test(bctx, "こんにちは、世界！", {101, 1655, 30217, 30194, 30188, 30198, 1635, 1745, 100, 1986, 102}); |
| 187 | +} |
0 commit comments