• State
  • Locked Locked
  • Replies 0 replies
  • Subscribers 99 subscribers
  • Views 167 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TDA4VM: Inference to handle float32 inputs and weights

Subraja P
Prodigy 80 points

Part Number: TDA4VM

Hi,

The demo applications in dl_demo from RTOS PSDK (9.1.0.6) has handled input datatype of UINT 8bit and 16bit.

I'm facing an error in conv2d.

main: src/tidl_conv2d_base.c:1433: int32_t TIDL_refConv2dProcess(const TIDL_CreateParams*, sTIDL_AlgLayer_t*, sTIDL_Layer_t*, sTIDL_ConvParams_t*, tidlConv2dBuffParams_t*, void*, void*): Assertion `0' failed.
Aborted (core dumped)

due to wrong tidlLayer->weightsElementSizeInBits value. 

tidlLayer->weightsElementSizeInBits is 8, but I expect 32 for float32 weights.

My custom model has input datatype and weights datatype as float32.

How to fix this issue ? 

Is there any sample inference available that handles float32 data ? 

Kindly let me know if you need more details.