Commits

Branch selector

 master

User selector

karpathy

Datepicker

All time

Commit History

Commits on Jun 22, 2023

• Merge pull request karpathy#309 from ho2103/master

  Show description for eba36e8
  karpathy

  authored
  eba36e8

Commits on Jun 18, 2023

• Merge pull request karpathy#305 from okuvshynov/fix_osx_dataload

  Show description for 4eb7a96
  karpathy

  authored
  4eb7a96

Commits on Jun 17, 2023

• Merge pull request karpathy#301 from okuvshynov/master

  Show description for 41d7014
  karpathy

  authored
  41d7014

Commits on Jun 14, 2023

• use WORLD_SIZE instead of device_count, supports both the case where the number of gpus we train on is smaller than gpus available, and also multinode training may be a bugfix

  karpathy

  committed
  7339b90

• Merge pull request karpathy#274 from apivovarov/gelu

  Show description for f08abb4
  karpathy

  authored
  f08abb4

• Merge pull request karpathy#275 from apivovarov/rm_unsqueeze

  Show description for 18ee6b6
  karpathy

  authored
  18ee6b6

• Merge pull request karpathy#270 from LaihoE/master

  Show description for ed7887c
  karpathy

  authored
  ed7887c

• Merge pull request karpathy#276 from apivovarov/gitign

  Show description for 8020bb5
  karpathy

  authored
  8020bb5

• Merge pull request karpathy#277 from apivovarov/is_bf16_supported

  Show description for 0f06d9b
  karpathy

  authored
  0f06d9b

• Merge pull request karpathy#286 from ctjlewis/master

  Show description for cf4835e
  karpathy

  authored
  cf4835e

Commits on May 6, 2023

• simplify configure_optimizers by a lot

  karpathy

  committed
  7fe4a09

Commits on Apr 18, 2023

• Merge pull request karpathy#247 from gnobre/macbook-run-instructions

  Show description for 196160b
  karpathy

  authored
  196160b

• Merge pull request karpathy#225 from otaviogood/grad_accum

  Show description for 21f9bff
  karpathy

  authored
  21f9bff

• Merge branch 'master' into grad_accum

  karpathy

  authored
  a6a708c

Commits on Apr 13, 2023

• Merge pull request karpathy#10 from LaihoE/master

  Show description for d9f4735
  karpathy

  authored
  d9f4735

• Merge pull request karpathy#146 from lutzroeder/master

  Show description for b288f4c
  karpathy

  authored
  b288f4c

• Merge pull request karpathy#74 from venusatuluri/fix_decode

  Show description for 079df20
  karpathy

  authored
  079df20

• Merge pull request karpathy#240 from YassineYousfi/master

  Show description for 01e48ec
  karpathy

  authored
  01e48ec

• Merge pull request karpathy#54 from MicroPanda123/luv

  Show description for 7840a66
  karpathy

  authored
  7840a66

• Merge pull request karpathy#128 from abrahamsangha/fix-typo

  Show description for 8abe215
  karpathy

  authored
  8abe215

• Merge pull request karpathy#142 from kovkev/patch-1

  Show description for ad62003
  karpathy

  authored
  ad62003

• Merge pull request karpathy#220 from python273/patch-1

  Show description for ea24604
  karpathy

  authored
  ea24604

• Merge pull request karpathy#224 from SnehalRaj/patch-1

  Show description for 8aeea6d
  karpathy

  authored
  8aeea6d

• Merge pull request karpathy#236 from ymurenko/master

  Show description for 2457471
  karpathy

  authored
  2457471

• fix minor bug where we have to scale the loss to account for gradient accumulation, which sums before backprop. note that this is not a major bug because AdamW is scale invariant. however, this did…

  Show description for 553f949
  karpathy

  committed
  553f949

Commits on Mar 12, 2023

• Merge pull request karpathy#199 from ChristianOrr/patch-1

  Show description for a82b33b
  karpathy

  authored
  a82b33b

Commits on Mar 7, 2023

• Merge pull request karpathy#195 from drisspg/enable_sdpa_with_nonzero_dropout

  Show description for 0d8fbd1
  karpathy

  authored
  0d8fbd1

Commits on Feb 15, 2023

• Merge pull request karpathy#145 from otaviogood/gradAccumStability

  Show description for ae3a8d5
  karpathy

  authored
  ae3a8d5

Commits on Feb 10, 2023

• fix misinformation in readme

  karpathy

  committed
  55c5069

Commits on Feb 7, 2023

• oops i should not be needing or multiplying by world_size to calculate mfu

  karpathy

  committed
  e58f0cf

Commits on Feb 6, 2023

• small tweaks, make default WD be 0.1 as is often cited, and remove spurious init of LayerNorm, which is already initialized at 1,0

  karpathy

  committed
  8b1e432

• bugfix we have to call the raw_model's estimate_mfu ty @jprobichaud for original PR

  karpathy

  committed
  ab21d6c

Commits on Feb 5, 2023

• also add a sampling/inference section

  karpathy

  committed
  f83dd03

• revamp the readme file to be a bit better and more accessible, i hope

  karpathy

  committed
  23a8e70

• tune the hyperparams a bit, in configs

  karpathy

  committed
  fce706c

Pagination

PreviousNext