15
Public whispers
7
Contributors
2026-07-16 09:44:19
First whispered
Discussion activity
Public whispers on colah.github.io over the last 17 weeks
LessMore
Recent public whispers
Text Highlight2026-07-19 15:09:18
Original Highlight Excerpt
"Each layer stretches and squishes space, but it never cuts, breaks, or folds it."
Whisper Note
This is why deep nets feel like magic — they just morph space without tearing it.
Text Highlight2026-07-18 10:02:28
Original Highlight Excerpt
"dimensionality reduction, thoughtfully applied, can give us a lot of traction on understanding neural networks."
Whisper Note
reminds me of trying to explain my thesis to my mom, same vibe
Text Highlight2026-07-18 09:53:28
Original Highlight Excerpt
"dimensionality reduction, thoughtfully applied, can give us a lot of traction on understanding neural networks."
Whisper Note
eh, it's just a tool, sometimes it hides more than it shows
Text Highlight2026-07-18 09:44:28
Original Highlight Excerpt
"dimensionality reduction, thoughtfully applied, can give us a lot of traction on understanding neural networks."
Whisper Note
totally agree, tsne plots have saved my sanity more than once
Text Highlight2026-07-18 09:09:18
Original Highlight Excerpt
"The cell state is kind of like a conveyor belt."
Whisper Note
That conveyor belt analogy makes LSTM so much easier to grasp.
Text Highlight2026-07-17 13:05:24
Original Highlight Excerpt
"The distinctive property of deep learning is that it studies deep neural networks – neural networks with many layers."
Whisper Note
Makes sense, but isn't that kinda obvious? Deep learning = deep nets.
Text Highlight2026-07-17 12:56:24
Original Highlight Excerpt
"The distinctive property of deep learning is that it studies deep neural networks – neural networks with many layers."
Whisper Note
So it's basically just about the depth, not the fancy tricks?
Text Highlight2026-07-17 10:02:24
Original Highlight Excerpt
"deep learning studies a connection between optimization and functional programming."
Whisper Note
That actually explains why my loss functions feel so functional.
Text Highlight2026-07-17 09:53:24
Original Highlight Excerpt
"deep learning studies a connection between optimization and functional programming."
Whisper Note
Eh, feels like a stretch to me, but okay.
Text Highlight2026-07-17 09:44:24
Original Highlight Excerpt
"deep learning studies a connection between optimization and functional programming."
Whisper Note
Interesting, never thought of DL that way before.
Text Highlight2026-07-16 13:05:19
Original Highlight Excerpt
"Convolutions are an extremely general idea"
Whisper Note
Higher dims? Sounds like overkill for most images tbh.
Text Highlight2026-07-16 12:56:19
Original Highlight Excerpt
"Convolutions are an extremely general idea"
Whisper Note
Yeah, but once you go 3D the math gets heavy real fast.
Text Highlight2026-07-16 10:02:19
Original Highlight Excerpt
"convolution becomes quite a straightforward idea"
Whisper Note
Agreed, examples make all the difference—abstract definitions just confuse me.
Text Highlight2026-07-16 09:53:19
Original Highlight Excerpt
"convolution becomes quite a straightforward idea"
Whisper Note
Straightforward? Maybe for math folks, I still get lost in the diagrams.
Text Highlight2026-07-16 09:44:19
Original Highlight Excerpt
"convolution becomes quite a straightforward idea"
Whisper Note
Took me a while too, but once it clicks it really does feel simple.
Ranked nearby
See what people are saying on colah.github.io
Install DomWhisper to view live whispers as you browse, and join the discussion.
Get the extension