Just to update this thread and correct myself. The book is absolutely fine. The problem was we were using it at bedtime with dimmed bedroom lighting. When we turn the bedroom lights back on, the book works fine again!
Is it that the reflection of the already-on LED (or maybe external light at the right angle) is hitting the sensor used for NOT and then creating a feedback loop to keep the light on when the switch isn't pressed? This happened a few times, and either changing the ambient light or holding the book tightly shut will resolve the issue on my copy.
edit: because the author is here, thanks for the book: an interesting idea and well-executed. Daughter doesn't mind the front page blue LED. I am more amused than annoyed: even a two-transistor circuit needs debugging sometimes.
Hey, this is my book! Thanks for sharing! I just posted an update today to all the backers about the the sequel for Big Babies book. I'm hoping to start shipping in the next couple weeks.
Just a quick note - I added both the books and checked out immediately from the two different urls in the comments and placed the order - before going to the home page. There was a discount option if you buy both the books together - which it did not apply at the checkout. So if you are buying both the books, do it from the homepage.
Great job on the product as well as the webpage!
Also, the signup flow is pretty painful. I have to upload a document, then pay, then go to my email and click on a link, that sends me another link to my email to sign in, then I complete the sign in, and it shows me the order information. And then no where does it let me chat with the robot about my doc...
That UX flow seems strange because the screen flow should just be: 1) upload, 2) pay, 3) chat. It shouldn't have asked you to check your email for anything.
Please feel free to email me at alfarez at gmail dot com if you're still seeing this issue.
As for handling large files, I'm using a popular method of breaking down the doc into small chunks and then just sending the relevant chunks to the AI.
How to pick the "relevant" chunks? Use embeddings and search for the embeddings that are "semantically closest" to the question being asked.
I think it's supposed to use the microphone to record your answers possibly? But it's not doing that for me either. I just get 15 seconds, and then "you failed."
It forced me to login, and I hate logging into stuff, so I didn't try it. But I like the idea. Another wrapper for GPT? Did you do any tweaking to the model? I think I've seen that developers can add training data to the model to give it more context or something"? Did you do any of that?
It was completely open originally, but I had to put it behind a login in order to limit how many times a user hit the OpenAI API.
I have creativity controls which adjust the "temperature", as well as revision controls which give you options to revise sections of the generated cover letter
Some feedback:
Firstly, this is loading full res photos on the homepage, which making the site extremely slow to load. You should generate some small thumbnails so that loading times aren't so terrible when looking at these gallery pages.
I would love more of a satirical approach. I would design it to be a very professional looking website, maybe knock off christie's or the louvre's design or something. Maybe have bio's about the authors (obviously keep it safe, because they're kids). And jack up the prices to be 100x what they currently are. Do really professional sounding descriptions for each piece (maybe you can use chatgpt to do this). "Art STUFF, is a still life Simeon. It's use of bright color pastel contradicts the subject matter of the piece and causes the viewer to consider the actual tools of the artist versus what the stereotypical tools that Simeon finds being pushed onto him." Or whatever...
Maybe professionally frame the pieces (or photoshop them to look professionally framed).
This seems like a fun side project. I'm also in the kid's market. I'd might be interested in partnering in some way if you're open to it.
