Method build_input_v_idxs

subprojects/llama.cpp/src/llama-kv-cache.cpp:1169–1183

}

ggml_tensor * llama_kv_cache::build_input_v_idxs(ggml_context * ctx, const llama_ubatch & ubatch) const {
    const uint32_t n_tokens = ubatch.n_tokens;

    ggml_tensor * v_idxs;

    if (!v_trans) {
        v_idxs = ggml_new_tensor_1d(ctx, GGML_TYPE_I64, n_tokens);
    } else {
        v_idxs = ggml_new_tensor_1d(ctx, GGML_TYPE_I64, n_tokens*hparams.n_embd_v_gqa_max());
    }

    ggml_set_input(v_idxs);

    return v_idxs;
}

void llama_kv_cache::set_input_k_idxs(ggml_tensor * dst, const llama_ubatch * ubatch, const slot_info & sinfo) const {
    const uint32_t n_tokens = ubatch->n_tokens;