Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 20, 2022

• tune description of the repo wrt references

  karpathy

  committed
  988aa59

Commits on Sep 15, 2022

• Merge pull request karpathy#3 from normanyu/fix-rnn-prev-state

  Show description for f61811b
  karpathy

  authored
  f61811b

• Fix bug in RNN where hprev always referred to start. Change so that hprev refers to output of previous cell

  

  Norman Yu

  authored and

  Norman Yu

  committed
  bf38625

Commits on Sep 2, 2022

• change readme

  karpathy

  committed
  2f5e8d7

Commits on Aug 22, 2022

• add a bag of words model that looks suspiciously similar to a transformer ;)

  karpathy

  committed
  c079e1c

• add an RNN and a GRU language model

  karpathy

  committed
  b697f43

• generalize makemore into other types of language models, and add bigram LM and an MLP LM

  karpathy

  committed
  6694b67

Commits on Aug 20, 2022

• fix comment

  karpathy

  committed
  50617fa

• implementation of InfiniteDataLoader sad

  karpathy

  committed
  d4ede45

• remove some guardrails for this simple of a use case

  karpathy

  committed
  a7c52cd

• remove gradient clipping i dont think its needed at this small scale

  karpathy

  committed
  4e0137d

• simplify optimizer init and delete code

  karpathy

  committed
  35435ec

• remove weight init, not needed at this scale

  karpathy

  committed
  d26d975

• add max steps

  karpathy

  committed
  0a19a59

Commits on Aug 19, 2022

• respect multigpu envs, e.g. cuda:2 designation should work

  karpathy

  committed
  055e7ee

• big refactor to make easier and api agree with mingpt more

  karpathy

  committed
  013af92

Commits on Jun 9, 2022

• add some generated examples of names for fun

  karpathy

  authored
  054568e

• split out train,test,new separately when reporting on sampling word identity

  karpathy

  authored
  c3aaadc

• small tweaks to support the Apple Silicon M1 chip device 'mps'. But this is not yet faster because a lot of ops are still being implemented pytorch/pytorch#77764 , in particular for us the layernor…

  Show description for e0a08f2
  karpathy

  committed
  e0a08f2

• first commit

  karpathy

  committed
  8f79bd0

• Initial commit

  karpathy

  authored
  180c4f7