I use the tool assets/tflite_export.py to convert effcientvit_b1 to tflite, I noticed that in the function TFLiteConverter, the parameter quantize_target_type is set to the default value 'uint8'. However, the output weight file does not include quantized and dequantized layers; each layer is still in float32 format."
I use the tool assets/tflite_export.py to convert effcientvit_b1 to tflite, I noticed that in the function TFLiteConverter, the parameter quantize_target_type is set to the default value 'uint8'. However, the output weight file does not include quantized and dequantized layers; each layer is still in float32 format."