This model is a 4-bit quantized version of the MNN model exported from Llama-3.2-1B-Instruct using llm-export.