Hi, I am completely new to TVM.
1.) One way of optimisation is to convert the whole pre trained model's graph in to an optimised one through TVM.([Like here on TVM tutorial ](https://tvm.apache.org/docs/how_to/compile_models/from_pytorch.html#sphx-glr-how-to-compile-models-from-pytorch-py) 2.) Another way of optimisation is to create your own layer and do custom optimisations ( [Like here](https://tvm.apache.org/docs/how_to/optimize_operators/opt_conv_cuda.html#sphx-glr-how-to-optimize-operators-opt-conv-cuda-py) ). Question a.) In this second layer case, in the tutorial I only see the speed test and not how to load pre trained weights.Which means if i create a custom layer in TVM (for which I already have the pre-trained weights in Pytorch) can I use the pre-trained weights from Pytorch in my custom layer in TVM? Or does that mean that if I create a convolution in TVM then I have to re-train the weights? Question b.) Can I customise optimisation for some of the layers by hand and auto tune for others in a pre-trained Pytorch model optimisation in 1? (Forgive me if the questions does not make sense, as I told, I am a complete newbie to TVM.) --- [Visit Topic](https://discuss.tvm.apache.org/t/optimising-pre-trained-models-with-custom-layer-implementation/12303/1) to respond. You are receiving this because you enabled mailing list mode. To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/085a4d0f596769fc80f60665a9ab55e85aeecf8bbe41363236024083821d3c43).
