@AndrejKarpathy
Hey. It must be some wierd bug with colab or the code.
Following your steps and running it in colab gives me several wierd issues. One issue i repeatedly get is:
hpr     See More | exact: False | approximate: False | maxdiff: 0.999999463558197
bngain | exact: False | approximate: False | maxdiff: 7.872222423553467
Don't know what to do here.    See Less

I almost completed Exercise 1 all on my own, but I had to step back for a day to refresh the basics because my college algebra was a bit rusty from 10 years of not using it. Exercises 2 and     See More overwhelmed me. However, when I follow your explanations, I understand everything. This is a huge because I remember that professors at my college couldn't explain complex concepts so easily. Andrej, you are a gift to this world! <3    See Less

For problems with exact values:
try casting tensor().to(torch.double)
Apparently divisions lose precision in pytorch     See Less

This was very insightful. Andrej you are the best!     See Less

finally made it to makemore pt.4 lfgggg     See Less

1:24:22 MY Vectorized version is below
I stopped the video, and from my first try have a vectorized version. I outdo     See More yey! 😂
X_expanded = F.one_hot(Xb, num_classes=C.shape[0]) * torch.ones([C.shape[1]] + list(Xb.shape) + [C.shape[0]])
X_expanded = X_expanded.movedim(0,-2)
X_expanded = X_expanded.movedim(-1,0)
r = (X_expanded * demb) # 27x32x3x10 * 32x3x10
r = r.sum((2,1)) # 27x10

It is not really efficient though, and not as numerically good, but good enough for educational purposes :)

C | exact: False | approximate: True | maxdiff: 3.725290298461914e-09

now I can go sleep and check what I can do better tomorrow ))    See Less

legend     See Less

Hi Andrej, could you explain why we need clone dnorm_logits for safety? Thanks!     See Less

simply the best! very good lessons with such maestry and passion, thanks a lot for sharing     See Less

Hey Andrej,

I don't know if you'll see this, but I just wanted to thank you whole heartedly for your awesome neural network playlist. It's by far the best and the most i     See More ntent on NNs I've ever come across. I really appreciate you sharing your knowledge for community. You're the best! Excited and awaiting for more such treasures!    See Less