You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Can you clarify what you mean by "show the NPU failing to load"?
From this line, it looks to me like you successfully loaded the model on NPU: 2024-05-14 19:55:39.273 9985-10016 tflite com.quicinc.imageclassification I Replacing 108 out of 108 node(s) with delegate (TfLiteQnnDelegate) node, yielding 1 partitions for the whole graph.
@kory I hope that the quantized model uses the HTP unit for operation, but it seems to use the cpu for operation.
2024-05-14 19:55:39.273 9985-10016 tflite com.quicinc.imageclassification I Replacing 108 out of 108 node(s) with delegate (TfLiteQnnDelegate) node, yielding 1 partitions for the whole graph.
2024-05-14 19:55:39.668 9985-10016 tflite com.quicinc.imageclassification I Created 0 GPU delegate kernels.
2024-05-14 19:55:39.688 9985-10016 tflite com.quicinc.imageclassification I Created TensorFlow Lite XNNPACK delegate for CPU.
2024-05-14 19:55:39.688 9985-10016 tflite com.quicinc.imageclassification I Replacing 71 out of 108 node(s) with delegate (TfLiteXNNPackDelegate) node, yielding 73 partitions for the whole graph.
Description:
I used the provided ImageClassification program and ran it on my Android device, which showed that the NPU failed to load.
Environment Details:
Chip: QCS6490
Android Subsystem: Android 13
QnnVersion: 2.28.2
Model: MobileNet-v2-Quantized (https://huggingface.co/qualcomm/MobileNet-v2-Quantized)
Model precision: int8
Log
The text was updated successfully, but these errors were encountered: