This model is a 4-bit quantized version of the MNN model exported from TinyLlama-1.1B-Chat using llm-export.