v1.0.0
v1.0.0
- Support fp32 on X86 AVX2 CPU
- Support partial fp32 operator(convolution, lstm) multi-threads parallel
- Support Tensorflow model
- Support more networks(Pointnet, ...)
- Support more networks int8 inference(TinyBert, NMT, ASR)
- Support time-series data acceleration
- Support Apple IOS phone