This model is a 4-bit quantized version of the MNN model exported from gte_sentence-embedding_multilingual-base using llm-export.