Commits

Breadcrumbs

History for

on98d60dd

User selector

All users

All time

Commit History

Commits on Dec 4, 2022

• test ops.split

  tiendung

  committed
  98d60dd

  Copy full SHA for 98d60dd

• try to fix lstm_cell hidden_size=2 input_size=1 batch_size=1

  tiendung

  committed
  19a1351

  Copy full SHA for 19a1351

• cleanup

  tiendung

  committed
  63c03f9

  Copy full SHA for 63c03f9

• refine

  tiendung

  committed
  f96d2b1

  Copy full SHA for f96d2b1

Commits on Dec 1, 2022

• refine

  tiendung

  committed
  663a7b9

  Copy full SHA for 663a7b9

Commits on Nov 29, 2022

• cleanup

  tiendung

  committed
  5c4d012

  Copy full SHA for 5c4d012

• fix

  tiendung

  committed
  3c1afe2

  Copy full SHA for 3c1afe2

Commits on Nov 27, 2022

• something fishy with how stack and split affect comp graph, it's hang because number of node is exponential with LSTM input sequence length, it runnable for length < 10

  tiendung

  committed
  324cfe0

  Copy full SHA for 324cfe0

Commits on Nov 26, 2022

• impl LSTM but fail tests and hang at backward

  tiendung

  committed
  868a77a

  Copy full SHA for 868a77a

Commits on Nov 24, 2022

• try to impl lstm cell, need to fix split op

  tiendung

  committed
  ef49607

  Copy full SHA for ef49607

• cleanup

  tiendung

  committed
  ab6699f

  Copy full SHA for ab6699f

Commits on Nov 23, 2022

• annotated rnn

  tiendung

  committed
  33decbc

  Copy full SHA for 33decbc

• fix

  tiendung

  committed
  5fbd58d

  Copy full SHA for 5fbd58d

• refine

  tiendung

  committed
  7a2aeaa

  Copy full SHA for 7a2aeaa

Commits on Nov 20, 2022

• refine

  tiendung

  committed
  7871834

  Copy full SHA for 7871834

• refine

  tiendung

  committed
  cfe3769

  Copy full SHA for cfe3769

• enhance train and test cifar10

  tiendung

  committed
  4197046

  Copy full SHA for 4197046

• use Flatten layer instead of hacking nn.Linear

  tiendung

  committed
  ff47ac0

  Copy full SHA for ff47ac0

• better test scripts

  tiendung

  committed
  5075886

  Copy full SHA for 5075886

• impl RNNCell

  tiendung

  committed
  73fd3b9

  Copy full SHA for 73fd3b9

Commits on Nov 19, 2022

• enhance test code

  tiendung

  committed
  a8b23d7

  Copy full SHA for a8b23d7

• fix nn.linear forward

  tiendung

  committed
  63989d7

  Copy full SHA for 63989d7

• refactor to test on various backends

  tiendung

  committed
  6d5cfa8

  Copy full SHA for 6d5cfa8

• revert triton matmul integration since many tests fail

  tiendung

  committed
  bbf78bc

  Copy full SHA for bbf78bc

• triton: integrate matmul kernel

  tiendung

  committed
  bf40965

  Copy full SHA for bf40965

• enhance test scripts

  tiendung

  committed
  3b03aaf

  Copy full SHA for 3b03aaf

• fix tensor tuple backward error

  tiendung

  committed
  fc60241

  Copy full SHA for fc60241

• move triton advanced kernels each to a separate file

  tiendung

  committed
  15bcd84

  Copy full SHA for 15bcd84

Commits on Nov 17, 2022

• use and test cuda_triton

  tiendung

  committed
  ce6d8e7

  Copy full SHA for ce6d8e7

Commits on Nov 16, 2022

• test across backends: cpu, cpu_numpy, cuda, and original numpy impls

  tiendung

  committed
  4b047ec

  Copy full SHA for 4b047ec

• try to impl resnet9

  tiendung

  committed
  4ae1280

  Copy full SHA for 4ae1280

• refine

  tiendung

  committed
  4a7a0f3

  Copy full SHA for 4a7a0f3

• enhance testing

  tiendung

  committed
  f5df334

  Copy full SHA for f5df334

• refine

  tiendung

  committed
  58687f1

  Copy full SHA for 58687f1

• try to fix oom due to tensor count keep growing every optim step

  tiendung

  committed
  225b97c

  Copy full SHA for 225b97c

Pagination

Previous

Next