Function llama_beam_search

llama.cpp:8917–8929

};

void llama_beam_search(llama_context * ctx,
                       llama_beam_search_callback_fn_t callback, void * callback_data,
                       size_t n_beams, int n_past, int n_predict) {
    assert(ctx);
    const int64_t t_start_sample_us = ggml_time_us();

    llama_beam_search_data beam_search_data(ctx, n_beams, n_past, n_predict);

    beam_search_data.loop(callback, callback_data);

    ctx->t_sample_us += ggml_time_us() - t_start_sample_us;
    ctx->n_sample++;
}

//
// quantization