Function main

tools/completion/completion.cpp:87–1003

#endif

int main(int argc, char ** argv) {
    std::setlocale(LC_NUMERIC, "C");

    common_params params;
    g_params = &params;

    common_init();

    if (!common_params_parse(argc, argv, params, LLAMA_EXAMPLE_COMPLETION, print_usage)) {
        return 1;
    }

    auto & sparams = params.sampling;

    // save choice to use color for later
    // (note for later: this is a slightly awkward choice)
    console::init(params.simple_io, params.use_color);
    atexit([]() { console::cleanup(); });

    if (params.embedding) {
        LOG_ERR("************\n");
        LOG_ERR("%s: please use the 'embedding' tool for embedding calculations\n", __func__);
        LOG_ERR("************\n\n");

        return 0;
    }

    if (params.n_ctx != 0 && params.n_ctx < 8) {
        LOG_WRN("%s: warning: minimum context size is 8, using minimum size.\n", __func__);
        params.n_ctx = 8;
    }

    if (params.rope_freq_base != 0.0) {
        LOG_WRN("%s: warning: changing RoPE frequency base to %g.\n", __func__, params.rope_freq_base);
    }

    if (params.rope_freq_scale != 0.0) {
        LOG_WRN("%s: warning: scaling RoPE frequency by %g.\n", __func__, params.rope_freq_scale);
    }

    LOG_INF("%s: llama backend init\n", __func__);

    llama_backend_init();
    llama_numa_init(params.numa);

    llama_model * model = nullptr;
    llama_context * ctx = nullptr;
    common_sampler * smpl = nullptr;

    g_model = &model;
    g_ctx = &ctx;
    g_smpl = &smpl;

    std::vector<common_chat_msg> chat_msgs;

    // load the model and apply lora adapter, if any
    LOG_INF("%s: load the model and apply lora adapter, if any\n", __func__);