Method build_inp_out_ids

subprojects/llama.cpp/src/llama-graph.cpp:1474–1493

}

ggml_tensor * llm_graph_context::build_inp_out_ids() const {
    // note: when all tokens are output, we could skip this optimization to spare the ggml_get_rows() calls,
    //       but this would make the graph topology depend on the number of output tokens, which can interere with
    //       features that require constant topology such as pipline parallelism
    //       ref: https://github.com/ggml-org/llama.cpp/pull/14275#issuecomment-2987424471
    //if (n_outputs < n_tokens) {
    //    return nullptr;
    //}

    auto inp = std::make_unique<llm_graph_input_out_ids>(hparams, cparams, n_outputs);

    auto & cur = inp->out_ids;

    cur = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_outputs);
    ggml_set_input(cur);

    res->add_input(std::move(inp));

    return cur;
}

ggml_tensor * llm_graph_context::build_inp_mean() const {
    auto inp = std::make_unique<llm_graph_input_mean>(cparams);