This model is a 4-bit quantized version of the MNN model exported from MobileLLM-600M using llm-export.