Since we built this book from the template, we should remove the unnecessary static file and other left overs from the template.

The ipynb doesnt seem to send any tensors to the GPU, however it calls does frequently call them to cpu for plotting etc.

Prof dev needs to be cleaned up. This page in particular https://neuroai.neuromatch.io/projects/professional_development/mentorship_program.html needs title cleaning

In this issue, we would like to mention all content contributors which are not mentioned yet.

`
16x16 images - Accuracy (RIMs): 74.29%
19x19 images - Accuracy (RIMs): 64.23%
24x24 images - Accuracy (RIMs): 34.96%
16x16 images - Accuracy (LSTMs): 72.68%
19x19 images - Accuracy (LSTMs): 44.21%
24x24 images - Accuracy (LSTMs): 15.17%

The accuracy of the model on 16x16 images is fairly close to what was observed on smaller images, indicating that the increase in size to 16x16 does not significantly impact the model's ability to recognize the images. However, RIMs demonstrate generalize better, when working with the larger 19x19 and 24x24 images - compared to LSTMs.
`

The accuracy with 16x16 is not fairly close to any other accuracy. The increase in size does impact performance negatively.

In the section titled Coding Exercise 3: Mixing Discrete Objects With Continuous Space, there is typo in the section. It says:

As you can seem, the similarity is destroyed, which is what we should expect.

But it should be:

As you can see, the similarity is destroyed, which is what we should expect.

Seem should be changed to see.

A heading is wrong.

In Section 2, of W1D2, https://neuroai.neuromatch.io/tutorials/W1D2_ComparingTasks/student/W1D2_Tutorial1.html#transfer-example-2-autoencoder-to-classification, the heading should be regression to autoencoder. Not what is written, i.e. "Autoencoder to Classification".

We need to double check our colab and kaggle links. I noticed that the intro page still uses the template for the colab links.

As we add duplicate information for both the processing and publish workflows, we should a setup workflow that is common for both and is used by both.

It looks like our all contributors config file is malformed.

In this section, a model with a shared workspace is implemented, but it's not trained or evaluated in any way. Feels like a section is missing.

The text for the tutorials needs to be proofread. It is pretty good as is, but it is often missing articles or is awkwardly worded.

Right now I don't think our environment is properly cached. Fixing this would speed up build time.

Every time we publish the notebooks, it takes a while to build. It better, if possible, to only build and publish the changed books. This would speed up the publish workflow significantly.

• Typos (3 occurrences of "futher" instead of "further")
• Line break missing right before "Going further: How learning rule-specific are the bias [...]"

Will create PR in just a moment.