The quantization config.json file is pushed first, followed by the quantized model weights.