This model is a 4-bit quantized version of the MNN model exported from MobileLLM-1B using llm-export.