[Review]: Introduction to deep learning

[svenvanderburg]

Lesson Title

Introduction to deep learning

Lesson Repository URL

https://github.com/carpentries-incubator/deep-learning-intro

Lesson Website URL

https://carpentries-incubator.github.io/deep-learning-intro/

Lesson Description

This is a hands-on introduction to the first steps in Deep Learning, intended for researchers who are familiar with (non-deep) Machine Learning.

The use of Deep Learning has seen a sharp increase of popularity and applicability over the last decade. While Deep Learning can be a useful tool for researchers from a wide range of domains, taking the first steps in the world of Deep Learning can be somewhat intimidating. This introduction aims to cover the basics of Deep Learning in a practical and hands-on manner, so that upon completion, you will be able to train your first neural network and understand what next steps to take to improve the model.

We start with explaining the basic concepts of neural networks, and then go through the different steps of a Deep Learning workflow. Learners will learn how to prepare data for deep learning, how to implement a basic Deep Learning model in Python with Keras, how to monitor and troubleshoot the training process and how to implement different layer types such as convolutional layers.

Author Usernames

@dsmits @psteinb @cpranav93 @colinsauze @CunliangGeng

Zenodo DOI

10.5281/zenodo.8308392

Differences From Existing Lessons

No response

Confirmation of Lesson Requirements

• [X] is the original work of the author(s), or that any content derived from another source is reused with permission and appropriate attribution
• [X] aligns with The Carpentries Code of Conduct
• [X] is published under a CC-BY or CC0 license
• [X] uses The Carpentries lesson template or Carpentries Workbench without significant customisation/adaptation.

JOSE Submission Requirements

• [X] the lesson repository includes paper.md and paper.bib files as described in the JOSE submission guide for learning modules

Potential Reviewers

No response

[svenvanderburg]

We're still running a final round of comments for the paper (see https://github.com/carpentries-incubator/deep-learning-intro/issues/364), the plan is to submit the paper on the 15th of September through: https://openjournals.readthedocs.io/en/jose/submitting.html#submitting-your-paper . I'm not sure how that relates to this review, are the 2 independent or should we wait for this review process to finish before submitting to JOSE?

[tobyhodges]

Great to see this submission, @svenvanderburg. As a listed contributor, I have a conflict of interest acting as editor for this one. I am going to find another community member who can fulfill the role for this review, and will post back here when it's ready.

To answer your question about review order: we have been asking lesson developers to submit the lesson for review here, before the JOSE review. See #11 and the related review in JOSE for an example.

[svenvanderburg]

@tobyhodges any update on finding an editor?

[svenvanderburg]

@tobyhodges ? 😇

[tobyhodges]

I have reached out to a potential guest editor for this review and am waiting for confirmation. Hoping to be able to follow up very soon!

[svenvanderburg]

I have reached out to a potential guest editor for this review and am waiting for confirmation. Hoping to be able to follow up very soon!

Perfect, thank you for the update 🙏

[tobyhodges]

Good news: @brownsarahm has kindly agreed to act as Guest Editor for this review. I am extremely grateful to her for being willing to take this on.

[brownsarahm]

I'll work on this in small bits, but this way it's all in one place and the authors could work on the (very minor) accessibility issues and one small note on setup that I have checked so far.

Editor Checklist - Intro to Deep Learning

Accessibility

• [x] All figures are also described in image alternative text or elsewhere in the lesson body.

• [x] The lesson uses appropriate heading levels:

  • [x] h2 is used for sections within a page.
  • [x] no “jumps” are present between heading levels e.g. h2->h4.
  • [x] no page contains more than one h1 element i.e. none of the source files include first-level headings.

• [x] The contrast ratio of text in all figures is at least 4.5:1.

• [x] the check boxes in the prereqs render poorly in the workbench + are an accessibility flag bc they're unlabeled "form elements" these should be removed

• [ ] listed browser support is not matched to jupyter browser support notably, Windows' current default browser is a chromium browser and therefore supported (and on general web compatibility, better than safari by a nontrivial margin) this should be updated to include more windows users without having to install

• most table headers are empty, but this might not be a problem

• in ep 2 there is a link to the sklearn docs with the text "here" this is rated "suspcicous" in accessibility terms, recommend "explained in the scikit-learn docs" instead of explained "here"

• minor syntax issue (: instead of =) causing "missing alt text" on an image in :

  • ep 1
  • ep 4 link to source

Content

• [x] The lesson teaches data and/or computational skills that could promote efficient, open, and reproducible research.
• [x] All exercises have solutions.
• [x] Opportunities for formative assessments are included and distributed throughout the lesson sufficiently to track learner progress. (We aim for at least one formative assessment every 10-15 minutes.)
• [x] Any data sets used in the lesson are published under a permissive open license i.e. CC0 or equivalent.

Datasets and licenses

• penguin in CC0
• weather is CC-BY
• CIFAR10 license is unclear/not easy to find/ may not exist, but it is public data

other content notes:

• deep learning workflow exercise in ep 1 solution is not a solution; there can be many answers here but maybe some details of what to check for in an answer would help
• "just" used dropna in clean missing values is not Carpentries language and can be read as dismissive. "ruin the training data" is also not very precise about the impact and does not acknowledge that there are different ways to handle missing data for reasons that can impact analyses. In line with teachign best practices, a slightly broader coverage/more intentional choice of using dropna, which is fine to do, would improve the lesson here
• ex reflecting on our resutls is oddly formatted and has an exercise in the solution
• solution to varying the dropout rate is a term that was introduced in a different form in a parenthetical in the previous episode, but not defined as it is used.

Design

• [x] Learning objectives are defined for the lesson and every episode.
• [x] The target audience of the lesson is identified specifically and in sufficient detail.

to fix:

• [x] question, objectives, and keypoints all display with "" around them

Repository

The lesson repository includes:

• [x] a CC-BY or CC0 license.
• [x] a CODE_OF_CONDUCT.md file that links to The Carpentries Code of Conduct.
• [x] a list of lesson maintainers.
• [x] tabs to display Issues and Pull Requests for the project.

• replace this with any further comments relating to the lesson repository.

Structure

• [x] Estimated times are included in every episode for teaching and completing exercises.
• [x] Episodes lengths are appropriate for the management of cognitive load throughout the lesson.

comment:

• all episodes are 55min + episode 3 is over 3 hours long?

Supporting information

The lesson includes:

• [x] a list of required prior skills and/or knowledge.
• [x] setup and installation instructions.
• [ ] a glossary of key terms or links out to definitions in an external glossary e.g. Glosario.

other setup note:

• it says "open a terminal" but that is insufficient information for a Windows user (they have multiple terminals and to launch jupyter only the "anaconda prompt" by default will work; CMD or powershell typically will not)

General

• replace this with any other comments that do not fit into any of the previous sections.

[svenvanderburg]

@brownsarahm any update on the progress?

[svenvanderburg]

@brownsarahm any update? Can you give us an indication when we can expect this to be done?

[brownsarahm]

Hi! Sorry, a bunch of unexpected things happened last fall, and then when you sent the first check-in I was off of work for the holidays. And the second came while I was in a deadline crunch,

This is now back in my active queue. I should finish the pre-reviewer stuff within a week and I'm looking for reviewers starting now.

[svenvanderburg]

Hi @brownsarahm. Cool, thanks for the update 🤗

[brownsarahm]

editorial checks are done and @tobyhodges and I are working on finding reviewers next.

The comment above has a few things for you all to look at now, but thanks for resolving all of the previously identified ones already!

[tobyhodges]

Thanks @brownsarahm. I have suggested a few reviewers to Sarah but if any of my fellow authors can also suggest anyone they think would be suitable, I am sure it would be helpful. (Please do not tag anyone here by their GitHub handle.)

[brownsarahm]

@svenvanderburg Do you have any updates in response to my final editorial check?

In particular, do you have responses to the concerns about:

• episode lengths
• dataset licenses

and ideally before we assign reviewers, it would be nice to resolve, but these are minor:

• " rendering on episode metadata
• typos breaking alt-text

[svenvanderburg]

Hey @brownsarahm.

Sorry, I totally missed that it was final! Thanks for extra pinging me @brownsarahm :)

Some answers:

Episode lengths

Indeed episode 3 takes a bit longer than the other ones, but not twice longer. In fact, the timing for the other episode is too optimistic: Here is a PR with more realistic timing. In comparison to other lessons the episodes are a bit longer, this is because we want to finish the full deep learning workflow in each episode. When teaching, this is not really a problem though, the workshop is actually pretty balanced in terms of cognitive load because in every episode we go through this deep learning workflow once, and conceptually it makes sense to have the cuts between episodes at these points. What do you think? We could maybe write this explanation in the introductory instructor notes?

Dataset licenses

If I understand correctly, the only problem is with the CIFAR-10 dataset. It is so widely used, but I never realized it doesn't actually have a license judging from the official website.

I found a paper that extracts the statement about citation as their license: Running example. In this paper, we download the CIFAR-10 dataset from its official website. Also on the CIFAR-10 website, we find the following request from the dataset creators: “Please cite (Krizhevsky et al., 2009) if you intend to use this dataset," alongside a link to the paper. We extract this as the dataset’s license.

It is actually a crawled dataset, and in the paper I don't read anything about the crawled images being under open-source license....

Anyway: do you think it is a problem that we use this dataset? It is such a central dataset in the field and so widely used. We would have to change the entire episode if we use a different dataset. We can write a comment about in the instructor notes.

Small issues

I resolved the two small issues you referred to, will be reviewed soon by one of my colleagues. We will soon pick up any other remaining issues.

• https://github.com/carpentries-incubator/deep-learning-intro/pull/444

Would be good to enter the next phase of reviewing. Let us know what we have to do to help this progress.

[brownsarahm]

Timing

Thanks for the explanation. I think at one level your strategy for putting breaks in the content makes sense. I am not certain if your claim about cognitive load is true, but also not certain that it is false.

However 3.5 hours without a break is a long time and most instructors will not want to give a break in the middle of an episode.

Maybe an instructor note reminding about breaks? (as context, I'm a maintainer on instructor training and we get a lot of complaints about not enough breaks there and we have them ~every 90 minutes).

Dataset License

In my understanding of carpentries policy a permissive license is required. Since the dataset was crawled, it likely has zero consent to be using the images from the original owners of the images. I think it probably does not have the same risks as imagenet, but should be checked. On the other hand, this is clearly, to me, an intended use by the people who curated the images into a dataset, despite them not putting a license on it and them possibly not having appropriate rights to the images either.

Whether it is okay or not is going to be up to Carpentries policy about the situation where there is no license. @tobyhodges can you help navigate that or let us know who else in the carpentries should be looped in?

[tobyhodges]

Thanks for tagging me @brownsarahm. I need to do a bit more reading and thinking about this, and will come back soon with a full response.

[tobyhodges]

Thanks for your patience while I took some time to read through the relevant pages and documents, and to reflect on the most appropriate course of action. I am sorry to say that I think we should replace the dataset in the lesson.

The lack of a license file in the dataset is somewhat problematic, even though the authors clearly intend for the dataset to be re-used and usage in the lesson is within the terms stated on their website. But my biggest concerns are with the unethical way in which the data was "collected." Images were scraped and modified for the dataset without any attempt at seeking permission from the copyright owners or giving them attribution, which feels unethical to me regardless of any arguments over its legality. (I am not a lawyer but it seems like the use may fall under "fair dealing" in Canadian copyright law, where the researchers who published the dataset are based.)

In Collaborative Lesson Development Training, alongside considerations of licensing, size, and complexity, we ask lesson developers to consider the ethics of the example datasets they include in their lessons. I would like to apply the same standard to lesson reviews in The Carpentries Lab.

I acknowledge that replacing the dataset will require significant new work on the part of the authors, and perhaps I should have noticed sooner and avoided some of this inconvenience. @svenvanderburg for my part, I would like to devote some time in the coming weeks to try to make the necessary changes (as I am already one of the authors). I hope I will be able to propose some alternative datasets soon, and of course it would help to have input from others with more DL experience than I have. However, please also be aware that you can withdraw the lesson from review here if you prefer.

Finally, many thanks to @brownsarahm for catching this and looping me into the discussion.

[svenvanderburg]

Timing

Ha, no 3,5 hour teaching without breaks is absurd! At the Netherlands eScience Center we usually teach in a schedule like [for this recently taught workshop](this https://esciencecenter-digital-skills.github.io/2024-02-05-ds-dl-intro/#schedule). Never more than 90 minutes of teaching! And I think beta pilots copied that schedule in rough lines. It doesn't matter that it is in the middle of an episode.

In addition, we swap instructors halfway the episodes which makes teaching load lighter as well.

See https://github.com/carpentries-incubator/deep-learning-intro/issues/446 for addressing this, do you agree @brownsarahm ? And thanks for bringing this up, this is a great outcome of the review. Since we always use the same schedule no matter what lesson material we use we had a blind spot here.

License

Thanks @tobyhodges for digging into the CIFAR10 license. I agree we should change it, indeed it goes against everything the Carpentries stands for...

So, the remaining issues to fix before the review are:

• https://github.com/carpentries-incubator/deep-learning-intro/issues/446
• https://github.com/carpentries-incubator/deep-learning-intro/issues/445

(and some more small comments from Sarah that we will definitely pick up the coming period but are not essential to do before the review)

Can you confirm this @brownsarahm ?

[brownsarahm]

Yes, this is correct, these two issues would get it to a point where it is ready for review.

[svenvanderburg]

@brownsarahm The two big remaining issues (#446 and #445) are resolved! Please proceed to the next step of the review process 🚀

[brownsarahm]

tiny, tiny errors introduced by the update:

• the index refers to the content as a workshop, but I think "lesson" is the more correct Carpentries term. the later use of "workshop" I think is correct.
• outlook refers to cifar-10 in a way that I think should be swapped to dollarstreet

tangent/option: the change to the dollar street data also opens a really great opportunity to do bias evaluation and talk about the importance of having many dimensions of diversity in a dataset. This is definitely out of scope for the reivew, but this dataset was published in a neurips paper showing some of this; the paper might be good to include in outlook as a reference (I learned about this dataset at a talk after I raised the concern here)

I will move forward on inviting reviewers!

[svenvanderburg]

Wow, you're so sharp @brownsarahm! We will address all your 3 comments/suggestions in: https://github.com/carpentries-incubator/deep-learning-intro/issues/462, https://github.com/carpentries-incubator/deep-learning-intro/issues/461 and https://github.com/carpentries-incubator/deep-learning-intro/issues/460.

I was already planning on using the dollarstreet dataset as example to open up a discussion on ethical AI next time we teach the lesson. I'm really happy that we use this dataset now.

Great, we're looking forward to the review!

[brownsarahm]

@likeajumprope thank you for volunteering to review lessons for The Carpentries Lab. Please can you confirm if you are happy to review this Introduction to Deep Learning lesson?

You can read more about the lesson review process in our Reviewer Guide.

[likeajumprope]

@likeajumprope thank you for volunteering to review lessons for The Carpentries Lab. Please can you confirm if you are happy to review this Introduction to Deep Learning lesson?

You can read more about the lesson review process in our Reviewer Guide.

Yes I am happy to accept the invitation for review.

[brownsarahm]

@mike-ivs thank you for volunteering to review lessons for The Carpentries Lab. Please can you confirm if you are happy to review this Introduction to Deep Learning lesson?

You can read more about the lesson review process in our Reviewer Guide.

[mike-ivs]

@mike-ivs thank you for volunteering to review lessons for The Carpentries Lab. Please can you confirm if you are happy to review this Introduction to Deep Learning lesson?

You can read more about the lesson review process in our Reviewer Guide.

Happy to review the lesson. We'll actually be teaching the beta lesson again next week!

[brownsarahm]

@svenvanderburg we have moved to the next phase!

I think we expect the reviews within about 6 weeks.

[mike-ivs]

Hi all,

I'm still working through the review and have currently gone through all of the supplementary material (instructor notes/glossary/references/etc) and the Summary&Setup, episode 1, and episode 2. I hope to get through the remaining episodes by the end of next week.

I'm pretty happy overall with the lesson, but will wait to post the "reviewer checklist" summary until i've finished all the episodes. In the meantime i'll post my comments/etc here so that there's something to get started with. And of course, some of the comments are suggestions/questions so feel free to answer as you wish!

Most of my comments are clarity/cognitive load related which is inevitable given the topic, but I do think the lesson does a stellar job of teaching Deep Learning already!

supplementary material (i.e. bonus material, instructor notes, references, etc)

• Overall i'm happy with the non-episode content.
• There are just a few small typos, wording tweaks, and a url fix (pesky non-static internet!) which I'll submit in a separate pull request and link back here (PR here).

Summary and Setup

• Good!
• We use a cloud environment when we teach this, and downloading+uploading the datasets is always fiddly, and so we use the Episode 3 instructor notes to download them in-line in Python. I think it's worth moving the download code(s) into a specific "Instructor callout" in the episodes themselves - the same way that breaks are mentioned.
• It might be worth exposing the "dataset download codes" to the learner view as well? Of course, there are pros and cons to selecting the "downloading on-the-fly" approach

Episode 1 - Introduction

Overview questions & objectives

• I think the Episode 1 & 2 "questions & objectives" may have gotten a bit mixed up at some point.
• Questions: "what is a neural network" should be in this episode instead of episode 2
• Objectives: "Identify the inputs and outputs of a deep neural network." should be moved to episode 2
• Questions and objectives could be re-ordered chronologically according to when they are covered in the episode (the other episodes do this I believe)

Figure fig/01_AI_ML_DL_differences.png

• The alt text of fig/01_AI_ML_DL_differences.png uses the acronyms AI/ML/DL/NN. NN is the only one that isn't defined in the episode, so worth defining somewhere, or just expanding out the acronyms.
• The cited source for fig/01_AI_ML_DL_differences.png doesn't seem to line up with that specific image. I think it might belong to Intel according to this source that cites it, the Intel Introduction to AI course Week 1 slides, and the colour palette compared to Intels other infographic style icons. But i'm not certain...
• Is it worth putting in a generative AI subset within the DL circle, like this example?

Activation functions

• Activation functions (and specifically ReLU) functions are first mentioned here but not really introduced.
• Shortly after there is an Activation Functions Challenge but there isn't really any episode content to prepare learners for this challenge.
• The acronym ReLU isn't defined anywhere, and the mathematical/programmatical definition is scattered through the lesson.
• I think Activation functions could benefit from a dedicated section in a similar way to how the Loss and optimisers are treated. This section could then provide the challenges for learners to work through. Like Loss and optimisers, this dedicated section could even be deferred to later episodes to ease the upfront cognitive load.
• Very minor point - we mention ReLU for the first time and provide it in a equation example but the neuron example just below that equation fig/01_neuron.png shows a sigmoid.

Neural network images

• Figure fig/01_xor_exercise.png in this exercise is the first time we see a NN with values embedded in the image, and so it might be worth including a bit more explanation to reduce the initial shock / cognitive load of understanding it.
• Similarly, figure fig/01_deep_network.png is another different way of looking at NN architecture and might be a bit heavy in terms of cognitive load (I appreciate DL/NNs are quite complex full stop!)

A few hyperlinks

• In the Examples of DL in research section: is it possible to find alternative non-paywall examples to replace the paywall examples?
• In the Installing Keras and other dependencies section: it looks like the setup url is broken.

Episode 2 - Classification by a neural network using Keras

Overview questions & objectives

• RECAP I think the Episode 1 & 2 "questions & objectives" may have gotten a bit mixed up at some point.
• RECAP Questions: "what is a neural network" should be in episode 1 instead of this episode
• RECAP Objectives: "Identify the inputs and outputs of a deep neural network." should be moved to this episode

Instructor note on the episode goal

• this note is really good at explaining "we will go through the full workflow once, and then go through later in greater detail"!
• Is it worth trying to emphasise this even more in the actual learner content of the lesson?

Palmer penguin links

• quite a few of these links are now outdated (and one now points to a spam site!). (PR here).

One hot encoding

• a question that was raised a few times when we delivered this lesson in past workshops was "why do we use 3 columns of 0/1 instead of just one column 0/1/2 for the 3 labels?" i.e. why one-hot instead of label encoding (without knowing the names)
• It might be worth elaborating on why we want to "make things more complex" via one-hot encoding i.e. discuss the "ranking issue" of label encoding.
• This may slow things down due to extra detail, but this was a point where we risked our learners falling behind without a proper explanation.

Random seeds

• it might be worth emphasising when and when not to use random seeds, in case learners simply "copy" what they learnt here?

Phrasing

• there are just a few words/sentences that could be tweaked - (PR here).

Chinstraps absent from confusion matrix

• Do we ever explain/explore why the chinstraps are not present in the the predictions in the confusion matrix? is it the specific random seed? the train/test has stratify/shuffle=True so the right steps were taken to avoid this issue... (short of a bad seed)
• It might be worth answering so that learners can see how to tackle the "black box" nature of DL/NNs. Leaving it unanswered or saying "a bad model" doesn't seem very satisfying or good practice.

General comments

• Is it worth putting acronyms+definitions in the glossary? There are a few.
• There is already an issue raised for this, but i noticed a few places where "workshop/classes" was used over "lesson/episode"

Once again, it's a nice lesson :)

[svenvanderburg]

@mike-ivs thank for your comments so far! Super useful. 🙏 Looking forward to the rest!

[likeajumprope]

Hi all,

Just to chime in: I am on it 😊

Feedback coming soon.

Best, Johanna

From: Sven van der Burg @.> Date: Tuesday, 9 July 2024 at 07:57 To: carpentries-lab/reviews @.> Cc: Johanna Bayer @.>, Mention @.> Subject: Re: [carpentries-lab/reviews] [Review]: Introduction to deep learning (Issue #25)

@mike-ivshttps://github.com/mike-ivs thank for your comments so far! Super useful. 🙏 Looking forward to the rest!

— Reply to this email directly, view it on GitHubhttps://github.com/carpentries-lab/reviews/issues/25#issuecomment-2216638067, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AFVBFNU5TDJ6YLAEHUVGGRLZLN3T5AVCNFSM6AAAAAA4HVHIKOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMJWGYZTQMBWG4. You are receiving this because you were mentioned.Message ID: @.***>

[brownsarahm]

Checking in @likeajumprope and @mike-ivs, could you each provide an updated ETA for your review this week(or the review itself if you happen to be done)?

[mike-ivs]

My apologies, life got in the way since my last post!

I've had a look at the changes so far (https://github.com/carpentries-incubator/deep-learning-intro/issues/482) and am very happy with them :) I'll submit my relevant link/typo PRs shortly. (PR here).

In terms of ETAs I aim to get the rest of the review finished up by the 2nd August at the latest, hopefully by the end of this week.

[svenvanderburg]

@mike-ivs no worries. Looking forward to the rest of your comments :)

[mike-ivs]

As promised, here's the remaining comments. I'll post the reviewer checklist after this along with overall comments/summary. Again, i've very happy with the lesson and most of my comments are aimed at improving the clarity / further reducing the cognitive-load of a fairly heavy topic! (the lesson does a very good job already).

Episode 3 - Monitor the training process

2) Identify inputs and outputs

• we don't explicitly mention what the outputs are here i.e. BASEL_sunshine for Day=i+1. It might transition things nicely into the data prepping section where we pull out the i+1 labels

4) Choose a pretrained model or start building architecture from scratch

• small typo in "function as a it proved"
• here we define the create_nn function
  • we pull X_data from the global scope rather than passing it in explicitly as an argument (which makes me feel uncomfortable!)
  • in the "Try to reduce the degree of overfitting" exercise we redefine create_nn but this time pass in node1 and node2 as arguments. We could reduce repetition by doing this the first time round?
  • in the "BatchNorm" section we redefine create_nn again. Is it worth giving this a different name for clarity/to avoid overriding the earlier model?

6) Train the model

• we introduce batch_size "for the first time" here, but we have already introduced it in the "Intermezzo:Batch gradient descent" section. I'd reword to say something like "As we discussed earlier"

9) Refine the model

• "Despite avoiding severe cases of overfitting," change to "In addition to avoiding severe cases of overfitting,"
• "Instead of comparing training runs for different number of epochs, early stopping allows to simply set the number of epochs to a desired maximum value." This bit confused me... early stopping would cause different numbers of epochs, wouldn't it? The max epochs is set in the model fitting, not by the early_stopping.

Episode 4: Advanced layer types

Dropout

• ""Let us add a dropout layer after each pooling layertowards the end of the network, that randomly drops 80% of the nodes.""
  • slight typo in "layertowards"
  • we only actually add one dropout after the final pooling in create_nn_with_dropout(), not after ever pooling like we said
  • slight inconsistency: in the "Challenge: Vary dropout rate" the solution DOES have a dropout layer after every pooling layer

pip install keras_tuner

• I realise it's in a potentially "watch only" section, but should we move this to pre-lesson installation instructions? Is there a good enough reason install part-way through? It's potentially an optional section, so I don't feel too strongly about it either way.

Episode 5: Transfer Learning

• it might be worth giving an overview (perhaps with a "conceptual" diagram) at the start of the episode of how we'll "wrap around" an existing model in order to use it. This might iron out the cognitive load of section 4.

2) Identify inputs and outputs

• we should at least mention them (the inputs/outputs), perhaps by simply referring to previous episode's work, and acknowledging that we'll cover them in the beginning of section 4

4) Choose a pre-trained model or start building architecture from scratch

• the beginning of section 4 encroaches a bit on the (lightweight) section 2 + 3.
  • i.e. the inputs are defined in section 4, and the upscale layer is a "kind of" data prep conceptually
  • it might be worth making it clearer earlier in the episode that steps 2+3 are somewhat dependant on the pretrained model. If we mention this is steps 2+3 this will help the workflow steps stick to their "conceptual lanes" a bit more (I appreciate that in the wild they merge together!). Section 4 is pretty chockablock with a bit of everything already so this could help reduce it's cognitive load by introducing things earlier.
• referring to top/head of the NN
  • it might not be clear what the "top/head" of an NN is when learning this: i.e. is it the first or last layer?
  • we build our NNs from bottom(input) to top(output), but read them top to bottom in the code/keras summary. The word "deep" also implies that the bottom is "further in" (I guess it's too late to call it TALL learning.)
  • A diagram would help clear this up by explaining where we add our own architecture wrapper around the existing model.
  • maybe a figure like this (source), maybe something even simpler

Outlook

• very minor, I'd bump instructor note to above the paragraph it references... just in case instructors need a reminder beforehand! (they all prep beforehand right??)

[mike-ivs]

Summary

Very happy with lesson overall and I would say it is pretty much ready to graduate beyond the incubator.

Quite a few of my comments are suggestions, and mostly aimed at improving clarity / reducing cognitive load on an inescapably concept-heavy topic.

The lesson contributors+maintainers+testers have done a great job!

Reviewer Checklist

Accessibility

• [x] The alternative text of all figures is accurate and sufficiently detailed *.
  • Large and/or complex figures may not be described completely in the alt text of the image and instead be described elsewhere in the main body of the episode.
• [x] The lesson content does not make extensive use of colloquialisms, region- or culture-specific references, or idioms.
• [x] The lesson content does not make extensive use of contractions (“can’t” instead of “cannot”, “we’ve” instead of “we have”, etc).

* To view the alternative text of an image, we recommend using the WAVE Web Accessibility Evaluation Tool or associated browser extensions. You can also inspect the source HTML of the image element in the developer tools of your web browser, or consult the source (R)Markdown file for the relevant page in the lesson repository on GitHub. For more information about what makes good alternative text for an image, read How to Design Great Alt Text: An Introduction, and Writing Alt Text for Data Visualization

Content

• The lesson content:
  • [x] conforms to The Carpentries Code of Conduct.
  • [x] meets the objectives defined by the authors.
  • [x] is appropriate for the target audience identified for the lesson.
  • [x] is accurate.
  • [x] is descriptive and easy to understand.
  • [x] is appropriately structured to manage cognitive load.
  • [x] does not use dismissive language.
• [x] Tools used in the lesson are open source or, where tools used are closed source/proprietary, there is a good reason for this e.g. no open source alternatives are available or widely-used in the lesson domain.
• [x] Any example data sets used in the lesson are accessible, well-described, available under a CC0 license, and representative of data typically encountered in the domain.
• [x] The lesson does not make use of superfluous data sets, e.g. increasing cognitive load for learners by introducing a new data set instead of reusing another that is already present in the lesson.
• [x] The example tasks and narrative of the lesson are appropriate and realistic.
• [x] The solutions to all exercises are accurate and sufficiently explained.
• [x] The lesson includes exercises in a variety of formats.
• [x] Exercise tasks and formats are appropriate for the expected experience level of the target audience.
• [x] All lesson and episode objectives are assessed by exercises or another opportunity for formative assessment.
• [x] Exercises are designed with diagnostic power.

Design

• [x] Learning objectives for the lesson and its episodes are clear, descriptive, and measurable. They focus on the skills being taught and not the functions/tools e.g. “filter the rows of a data frame based on the contents of one or more columns,” rather than “use the filter function on a data frame.”
• [x] The target audience identified for the lesson is specific and realistic.

Supporting information

• [x] The list of required prior skills and/or knowledge is complete and accurate.
• [x] The setup and installation instructions are complete, accurate, and easy to follow.
• [x] No key terms are missing from the lesson glossary or are not linked to definitions in an external glossary e.g. Glosario.

[svenvanderburg]

Great! @mike-ivs thank you for your review! 🙏

[brownsarahm]

Hi @likeajumprope checking in for n ETA on your review