Function main

examples/perplexity/perplexity.cpp:680–750

}

int main(int argc, char ** argv) {
    gpt_params params;

    params.n_batch = 512;
    if (!gpt_params_parse(argc, argv, params)) {
        return 1;
    }

    params.logits_all = true;
    params.n_batch = std::min(params.n_batch, params.n_ctx);

    if (params.ppl_stride > 0) {
        fprintf(stderr, "Will perform strided perplexity calculation -> adjusting context size from %d to %d\n",
                params.n_ctx, params.n_ctx + params.ppl_stride/2);
        params.n_ctx += params.ppl_stride/2;
    }

    print_build_info();

    if (params.seed == LLAMA_DEFAULT_SEED) {
        params.seed = time(NULL);
    }

    fprintf(stderr, "%s: seed  = %u\n", __func__, params.seed);

    std::mt19937 rng(params.seed);
    if (params.random_prompt) {
        params.prompt = gpt_random_prompt(rng);
    }

    llama_backend_init(params.numa);

    llama_model * model;
    llama_context * ctx;

    // load the model and apply lora adapter, if any
    std::tie(model, ctx) = llama_init_from_gpt_params(params);
    if (model == NULL) {
        fprintf(stderr, "%s: error: unable to load model\n", __func__);
        return 1;
    }

    const int n_ctx_train = llama_n_ctx_train(model);
    if (params.n_ctx > n_ctx_train) {
        fprintf(stderr, "%s: warning: model was trained on only %d context tokens (%d specified)\n",
                __func__, n_ctx_train, params.n_ctx);
    }

    // print system information
    {
        fprintf(stderr, "\n");
        fprintf(stderr, "%s\n", get_system_info(params).c_str());
    }

    struct results_perplexity results;
    if (params.hellaswag) {
        hellaswag_score(ctx, params);
    } else {