Commits

Branch selector

 shapetracker

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 28, 2023

• More comprehensive merge_views

  corsix

  committed
  c79f85a

  Copy full SHA for c79f85a

Commits on Dec 27, 2023

• cleanup float4 grouping in global_load and global_store (#2942)

  Show description for 50f2e31
  chenyuxyz

  authored
  50f2e31

  Copy full SHA for 50f2e31

Commits on Dec 26, 2023

• minor cleanup in kernel and linearizer (#2937)

  Show description for 54629b5
  chenyuxyz

  authored
  54629b5

  Copy full SHA for 54629b5

• fix PADTO optimization (#2935)

  Show description for 820f2e0
  chenyuxyz

  authored
  820f2e0

  Copy full SHA for 820f2e0

Commits on Dec 25, 2023

• tensor == tensor should be bool (#2916)

  Show description for dca5e4f
  Qazalin

  authored
  dca5e4f

  Copy full SHA for dca5e4f

• test dtypes of return values of cumsum, argmax/min, multinomial (#2933)

  Show description for 8a8aed2
  chenyuxyz

  authored
  8a8aed2

  Copy full SHA for 8a8aed2

• green linearizer asserts for ops (#2800)

  Show description for 12996d3
  

  Qazalin

  and

  geohot

  authored
  12996d3

  Copy full SHA for 12996d3

• hotfix fix coder. RMSNorm cannot have float16 input (#2932)

  Show description for 1fb815e
  chenyuxyz

  authored
  1fb815e

  Copy full SHA for 1fb815e

• add CMPEQ (#2931)

  Show description for b469fe3
  chenyuxyz

  authored
  b469fe3

  Copy full SHA for b469fe3

• Fixed Tensor.randint() not accepting tuple shapes (#2923)

  Show description for 016aebc
  BreakTheBeta

  authored
  016aebc

  Copy full SHA for 016aebc

Commits on Dec 24, 2023

• clean up manual cast dtypes in tensor.py (#2930)

  Show description for 2dc99af
  chenyuxyz

  authored
  2dc99af

  Copy full SHA for 2dc99af

• Einsum space fix (#2927)

  Show description for 8de1fc2
  Isalia20

  authored
  8de1fc2

  Copy full SHA for 8de1fc2

• use at least int32 and uint32 for sum output (#2926)

  Show description for b55b55d
  chenyuxyz

  authored
  b55b55d

  Copy full SHA for b55b55d

Commits on Dec 23, 2023

• tensor.py cleanup around Tensor.slice (#2921)

  Show description for d424bab
  chenyuxyz

  authored
  d424bab

  Copy full SHA for d424bab

Commits on Dec 22, 2023

• cleanup test_dtype_alu (#2919)

  Show description for 089703a
  chenyuxyz

  authored
  089703a

  Copy full SHA for 089703a

• less outdated abstraction.py (#2917)

  Show description for 3ba591c
  chenyuxyz

  authored
  3ba591c

  Copy full SHA for 3ba591c

• s/lazydata.realized/lazydata.base.realized/g (#2914)

  Show description for 50927de
  chenyuxyz

  authored
  50927de

  Copy full SHA for 50927de

• bugfix Tensor.item when it's unbased (#2913)

  Show description for 2783e1b
  chenyuxyz

  authored
  2783e1b

  Copy full SHA for 2783e1b

• Disk shm refactor (#2912)

  Show description for c3133ad
  SkullMag

  authored
  c3133ad

  Copy full SHA for c3133ad

• don't use numpy to create Tensor(None) (#2909)

  Show description for 3855432
  chenyuxyz

  authored
  3855432

  Copy full SHA for 3855432

• update onnx model links (#2908)

  Show description for 50cfb1f
  chenyuxyz

  authored
  50cfb1f

  Copy full SHA for 50cfb1f

• remove the different rtol / atol for openpilot CUDA in benchmark (#2907)

  Show description for 1bbeb3f
  chenyuxyz

  authored
  1bbeb3f

  Copy full SHA for 1bbeb3f

• fuzz default dtypes for some test_dtype tests (#2906)

  Show description for a543d8b
  chenyuxyz

  authored
  a543d8b

  Copy full SHA for a543d8b

• catch cycles in print_tree (#2891)

  Show description for 5f3d5cf
  wozeparrot

  authored
  5f3d5cf

  Copy full SHA for 5f3d5cf

• minor cleanups / remove that op (#2905)

  geohot

  authored
  4432cb1

  Copy full SHA for 4432cb1

• onnx_ops formatting cleanup (#2904)

  Show description for fd0ba33
  chenyuxyz

  authored
  fd0ba33

  Copy full SHA for fd0ba33

Commits on Dec 21, 2023

• apply the multitensor optimizations in lazy.py (#2901)

  Show description for 5cac633
  geohot

  authored
  5cac633

  Copy full SHA for 5cac633

• reenable one onnx test failed due to dtype (#2902)

  chenyuxyz

  authored
  5bf43c9

  Copy full SHA for 5bf43c9

• use np.less and torch.lt for CMPLT (#2899)

  Show description for 677ae76
  chenyuxyz

  authored
  677ae76

  Copy full SHA for 677ae76

• render_locals takes a dtype (#2873)

  Show description for d2e9245
  

  Qazalin

  and

  chenyuxyz

  authored
  d2e9245

  Copy full SHA for d2e9245

• don't match dtype with first input in where (#2898)

  Show description for 6116039
  chenyuxyz

  authored
  6116039

  Copy full SHA for 6116039

• increase stable diffusion validation threshold 1e-4 -> 3e-4 (#2897)

  Show description for 7dc3352
  chenyuxyz

  authored
  7dc3352

  Copy full SHA for 7dc3352

• Move the webgpu CMPLT hack to one place (#2895)

  Show description for 24e79e0
  Qazalin

  authored
  24e79e0

  Copy full SHA for 24e79e0

• fix readme typo

  geohot

  committed
  852ef57

  Copy full SHA for 852ef57

• hotfix: compare on ids

  geohot

  committed
  193109a

  Copy full SHA for 193109a

Pagination

Previous

Next