Function main

subprojects/llama.cpp/tools/cli/cli.cpp:189–421

};

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

    params.verbosity = LOG_LEVEL_ERROR; // by default, less verbose logs

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

    // TODO: maybe support it later?
    if (params.conversation_mode == COMMON_CONVERSATION_MODE_DISABLED) {
        console::error("--no-conversation is not supported by llama-cli\n");
        console::error("please use llama-completion instead\n");
    }

    common_init();

    // struct that contains llama context and inference
    cli_context ctx_cli(params);

    llama_backend_init();
    llama_numa_init(params.numa);

    // TODO: avoid using atexit() here by making `console` a singleton
    console::init(params.simple_io, params.use_color);
    atexit([]() { console::cleanup(); });

    console::set_display(DISPLAY_TYPE_RESET);

#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
    struct sigaction sigint_action;
    sigint_action.sa_handler = signal_handler;
    sigemptyset (&sigint_action.sa_mask);
    sigint_action.sa_flags = 0;
    sigaction(SIGINT, &sigint_action, NULL);
    sigaction(SIGTERM, &sigint_action, NULL);
#elif defined (_WIN32)
    auto console_ctrl_handler = +[](DWORD ctrl_type) -> BOOL {
        return (ctrl_type == CTRL_C_EVENT) ? (signal_handler(SIGINT), true) : false;
    };
    SetConsoleCtrlHandler(reinterpret_cast<PHANDLER_ROUTINE>(console_ctrl_handler), true);
#endif

    console::log("\nLoading model... "); // followed by loading animation
    console::spinner::start();
    if (!ctx_cli.ctx_server.load_model(params)) {
        console::spinner::stop();
        console::error("\nFailed to load the model\n");
        return 1;
    }

    console::spinner::stop();
    console::log("\n");

    std::thread inference_thread([&ctx_cli]() {
        ctx_cli.ctx_server.start_loop();
    });