karpathy/nn-zero-to-hero

Issues

• Bug on `lectures/micrograd/micrograd_lecture_second_half_roughly.ipynb`

  #52 opened 2 months ago by younes-io
  1

• PME _ 3 Issue 2

  #65 opened 2 months ago by Hrushi0d
  0

• PME-3-Issue

  #64 opened 2 months ago by Hrushi0d
  0

• flash attention: simplified

  #61 opened 3 months ago
  2

• makemore/lecture4/backprop/dC

  #57 opened 6 months ago by we89
  2

• Fun solution to loop replacement (`makemore_part4_backprop.ipynb`)

  #46 opened 9 months ago by jchwenger
  1

• a new way of backprob of C[Xb] instead of for loop

  #55 opened 6 months ago by haduoken
  1

• No keepdim in dbnmeani calculation in makemore_part4_backprop.ipynb

  #59 opened 6 months ago by troeshust
  0

• Question on Simple Optimization Step

  #54 opened 7 months ago by mitnk
  2

• Neural Network

  #53 opened 7 months ago by Milad1DataScientist
  0

• makemore_part4_backprop dhpreact exact part is False.

  #45 opened 10 months ago by Stealeristaken
  2

• Problem with makemore_part2_mlp.ipynb

  #50 opened 8 months ago by dewijones92
  1

• name.txt is missing

  #48 opened 8 months ago by patricxu
  1

• Numerical instability in Google Colab - Part 4 of Makemore

  #13 opened 2 years ago by sachag678
  7

• BatchNorm1d in makemore part3 may have error

  #44 opened 10 months ago by LYK-love
  0

• need Discussions :)

  #42 opened a year ago by fenneishi
  0

• TypeError: unsupported operand type(s) for +: 'int' and 'Value'

  #41 opened a year ago by EruditeBoen
  1

• Error in makemore_part1_bigrams.ipynb

  #37 opened 2 years ago by mksamelson
  1

• Missing L.backward() before constructing graph

  #40 opened a year ago by arun477
  1

• zero_grad implementation correctness question

  #30 opened 2 years ago by zivanfi
  3

• Shouldn't names.txt be added to this repo?

  #16 opened 2 years ago by delbarital
  1

• Factorized code for dC in makemore_part4_backprop.ipynb

  #29 opened 2 years ago by bzkhalil
  1

• RuntimeError: Trying to backward through the graph a second time (or directly access saved tensors after they have already been freed). Saved intermediate values of the graph are freed when you call .backward() or autograd.grad(). Specify retain_graph=True if you need to backward through the graph a second time or if you need to access saved tensors after calling backward.

  #31 opened 2 years ago by Bie401
  1

• Not bug, just curiosity

  #8 opened 2 years ago by ax-anoop
  3

• Discrepancy in makemore_part1_bigrams single sample draw

  #34 opened 2 years ago by anoopelias
  2

• Suggestion: Time Series Forecasting and Transfer Learning

  #25 opened 2 years ago by fdelcab
  0

• Zero to hero

  #24 opened 2 years ago by feizalsaid
  0

• If I place more than 3 layers of FlattenConsecutive in makemore_part5_cnn1.ipynb, I get error: not enough values to unpack (expected 3, got 2)

  #23 opened 2 years ago by NarenZen
  0

• difference on calculating loss for count-matrix

  #18 opened 2 years ago by M-Amrollahi
  0

• intro to neural networks first

  #15 opened 2 years ago by culurciello
  3

• Not a real issue: Video links broken

  #14 opened 2 years ago by anshul360
  5

• b

  #2 opened 2 years ago by harpreetmann24
  1

• Bug in makemore part-1, in computing the liklihood

  #1 opened 2 years ago by Majdoddin
  1