As seen in the following graph, converting a model from TF into a quantized TFLite model greatly decreases the inference time and latency of a model:.
確定! 回上一頁