const int64_t t_main_start_us = ggml_time_us();

bert_params params;

params.model = "../../models/all-MiniLM-L6-v2/ggml-model-f32.bin";

if (bert_params_parse(argc, argv, params) == false) {

return 1;

}

int64_t t_load_us = 0;

bert_ctx * bctx;

// load the model

{

const int64_t t_start_us = ggml_time_us();

if ((bctx = bert_load_from_file(params.model)) == nullptr) {

fprintf(stderr, "%s: failed to load model from '%s'\n", __func__, params.model);

return 1;

}

t_load_us = ggml_time_us() - t_start_us;

}

int64_t t_eval_us = 0;

int64_t t_start_us = ggml_time_us();

int N = bert_n_max_tokens(bctx);

// tokenize the prompt

std::vector<bert_vocab_id> tokens(N);

int n_tokens;

bert_tokenize(bctx, params.prompt, tokens.data(), &n_tokens, N);

tokens.resize(n_tokens);

printf("%s: number of tokens in prompt = %zu\n", __func__, tokens.size());

printf("\n");

printf("[");

for (auto& tok : tokens) {

printf("%d, ", tok);

}

printf("]\n");

for (auto& tok : tokens) {

printf("%d -> %s\n", tok, bert_vocab_id_to_token(bctx, tok));

}

std::vector<float> embeddings(bert_n_embd(bctx));

bert_eval(bctx, params.n_threads, tokens.data(), n_tokens, embeddings.data());

t_eval_us += ggml_time_us() - t_start_us;

printf("[");

for(auto e : embeddings) {

printf("%1.4f, ", e);

}

printf("]\n");

// report timing

{

const int64_t t_main_end_us = ggml_time_us();

printf("\n\n");

//printf("%s: mem per token = %8zu bytes\n", __func__, mem_per_token);

printf("%s: load time = %8.2f ms\n", __func__, t_load_us/1000.0f);

printf("%s: eval time = %8.2f ms / %.2f ms per token\n", __func__, t_eval_us/1000.0f, t_eval_us/1000.0f/tokens.size());

printf("%s: total time = %8.2f ms\n", __func__, (t_main_end_us - t_main_start_us)/1000.0f);

}

return 0;

}