Getting Started - Neural Network Inference

d2l
#1

A placeholder for the book, discussion threads are embedded into the following page:

http://tvm.d2l.ai/chapter_getting_started/from_mxnet.html

2 Likes
#2

I consider the readers will confuse about Relay as we never mentioned it before in this book. Should we have a chapter to explain what is Relay and how it works with TVM?

1 Like
#3

Agree, plan to have a section late (in the neural network chapter) to explain it in details. Also it’s a good idea to explain a little bit more about relay in this section.

#4

Thanks for this tutorial!

#6

Thanks for your comprehensive tutorial! However, there might be a minor typo in the end of chapter 1.3.5. As you can see, the code exactly above section 1.3.6 which is tvm.testing.assert_allclose(loaded_scores, scores) should be np.testing.assert_allclose(loaded_scores, scores) .

#7

Thanks for this tutorial! I want to recommend 2 additional stuffs that I got suffered while reading this tutorial.

  1. Simple instruction about where to download ImageNet 1k dataset.
  2. Additional comments about image_processing function.

Thank you.

3 Likes
#8

Thanks for the tutorial! But it’s better to have command to download related files: …/data/imagenet1k_labels.txt …/data/cat.jpg