This model is a 4-bit quantized version of the MNN model exported from OpenELM-270M-Instruct using llm-export.