Function bark_model_quantize

bark.cpp:2354–2431

}

bool bark_model_quantize(const char* fname_inp, const char* fname_out, enum ggml_ftype ftype) {
    printf("%s: loading model from '%s'\n", __func__, fname_inp);

    std::string fname_inp_str(fname_inp);
    std::string fname_out_str(fname_out);

    auto fin = std::ifstream(fname_inp_str, std::ios::binary);
    if (!fin) {
        fprintf(stderr, "%s: failed to open '%s' for reading\n", __func__, fname_inp);
        return false;
    }

    auto fout = std::ofstream(fname_out_str, std::ios::binary);
    if (!fout) {
        fprintf(stderr, "%s: failed to open '%s' for writing\n", __func__, fname_out);
        return false;
    }

    // verify magic
    {
        uint32_t magic;
        fin.read((char*)&magic, sizeof(magic));
        if (magic != GGML_FILE_MAGIC) {
            fprintf(stderr, "%s: invalid model file '%s' (bad magic)\n", __func__, fname_inp);
            return false;
        }

        fout.write((char*)&magic, sizeof(magic));
    }

    // transfer vocab data from fin to fout
    {
        uint32_t n_vocab;
        read_safe(fin, n_vocab);
        write_safe(fout, n_vocab);

        std::string word;
        for (int i = 0; i < n_vocab; i++) {
            uint32_t len;
            read_safe(fin, len);
            write_safe(fout, len);

            word.resize(len);
            fin.read((char*)word.data(), len);
            fout.write((char*)word.data(), len);
        }
    }

    // text model
    if (!bark_model_weights_quantize(fin, fout, ftype)) {
        fprintf(stderr, "%s: failed to quantize text model\n", __func__);
        return false;
    }

    // coarse model
    if (!bark_model_weights_quantize(fin, fout, ftype)) {
        fprintf(stderr, "%s: failed to quantize coarse model\n", __func__);
        return false;