Arnaud de Gérard

ONNX inference does not work with CUDA 12

1

Exception occurs when upscaling an image with onnxruntime_gpu if execution provider is either CUDA or TensorRT. This is a known issue because CUDA 12 it is not yet supported. [Requirements...

NCNN optimizer: wrong assignment/test will corrupt a layer

6

I was doing some code review... (main branch) https://github.com/chaiNNer-org/chaiNNer/blob/9dac4fc7b6bf9934cf8a009c6c7b456990fef827/backend/src/nodes/impl/ncnn/optimizer.py#L658 ```python def __eliminate_dropout .... .... j = i - 1 for j in range(i - 1, -1, -1): ... else: j...