TDA4VM: TIDL inference indata格式疑问

version: /ti-processor-sdk-rtos-j721e-evm-07_01_00_11/tidl_j7_01_03_00_11/

我原来训练好的模型输入是一个480行640列的张量，元素为取值0.0～1.0之间的float,
example:
shape(480.640)
[[0.98039216 0.98039216 0.98039216 ... 0.57254905 0.5803922  0.5882353 ]
 [0.98039216 0.98039216 0.98039216 ... 0.5803922  0.5882353  0.59607846]
 [0.98039216 0.98039216 0.98039216 ... 0.5882353  0.59607846 0.6117647 ]
 ...
 [0.5529412  0.53333336 0.5254902  ... 0.25882354 0.25490198 0.2509804 ]
 [0.5411765  0.5411765  0.53333336 ... 0.25490198 0.25882354 0.2627451 ]
 [0.5529412  0.5686275  0.56078434 ... 0.2509804  0.2627451  0.27450982]]

请问我的indata文件.y应该是怎么样的编码格式？

config:
modelType          = 2
numParamBits = 8
numFeatureBits = 8
inputNetFile      = "/home/swl/ws/tidl/superpoint_v1.onnx"
outputNetFile      = "/home/swl/ws/tidl/int8_model/tidl_net_superpoint_net1.1.bin"
outputParamsFile   = "/home/swl/ws/tidl/int8_model/tidl_io_superpoint_net1.1_"
inWidth  = 640
inHeight = 480 
inNumChannels = 1
perfSimConfig  = "/home/swl/ti-processor-sdk-rtos-j721e-evm-07_01_00_11/tidl_j7_01_03_00_11/ti_dl/test/testvecs/config/import/device_config.cfg"
debugTraceLevel = 3
writeTraceLevel = 0
inFileFormat    = 1
inElementType = 0
rawDataInElementType = 0
postProcType = 0
numFrames   = 1
netBinFile      = "/home/swl/ws/tidl/int8_model/tidl_net_superpoint_net1.1.bin"
ioConfigFile    = "/home/swl/ws/tidl/int8_model/tidl_io_superpoint_net1.1_1.bin"
inData  =   "/home/swl/ws/tidl/test_int8_in.y"
outData =   "/home/swl/ws/tidl/int8_model/out/out.bin"