Skip to content

Commit 2505639

Browse files
vkuzovkuzo
committed
Update
[ghstack-poisoned]
2 parents e228991 + 17e7e77 commit 2505639

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

41 files changed

+2345
-620
lines changed

‎.github/workflows/doc_build.yml

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -9,6 +9,9 @@ on:
99
tags:
1010
- v[0-9]+.[0-9]+.[0-9]
1111
- v[0-9]+.[0-9]+.[0-9]+-rc[0-9]+
12+
paths:
13+
- 'docs/**'
14+
- '!docs/**'
1215
pull_request:
1316
workflow_dispatch:
1417

‎.github/workflows/float8_test.yml

Lines changed: 7 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -25,9 +25,14 @@ jobs:
2525
include:
2626
- name: SM-89
2727
runs-on: linux.g6.4xlarge.experimental.nvidia.gpu
28-
torch-spec: '--pre torch --index-url https://download.pytorch.org/whl/nightly/cu121'
28+
torch-spec: '--pre torch==2.7.0.dev20250122 --index-url https://download.pytorch.org/whl/nightly/cu124'
2929
gpu-arch-type: "cuda"
30-
gpu-arch-version: "12.1"
30+
gpu-arch-version: "12.4"
31+
- name: H100
32+
runs-on: linux.aws.h100
33+
torch-spec: '--pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu124'
34+
gpu-arch-type: "cuda"
35+
gpu-arch-version: "12.4"
3136

3237
permissions:
3338
id-token: write

‎.github/workflows/float8nocompile_test.yaml

Lines changed: 55 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,55 @@
1+
name: Run Float8nocompile Tests
2+
3+
on:
4+
push:
5+
branches:
6+
- main
7+
- 'gh/**'
8+
paths:
9+
- 'torchao/prototype/float8nocompile/**'
10+
- '!torchao/prototype/float8nocompile/**'
11+
pull_request:
12+
branches:
13+
- main
14+
- 'gh/**'
15+
paths:
16+
- 'torchao/prototype/float8nocompile/**'
17+
- '!torchao/prototype/float8nocompile/**'
18+
19+
concurrency:
20+
group: floatnocompile_test-${{ github.workflow }}-${{ github.ref == 'refs/heads/main' && github.run_number || github.ref }}
21+
cancel-in-progress: true
22+
23+
env:
24+
HF_TOKEN: ${{ secrets.HF_TOKEN }}
25+
26+
jobs:
27+
test:
28+
strategy:
29+
fail-fast: false
30+
matrix:
31+
include:
32+
- name: SM-89
33+
runs-on: linux.g6.4xlarge.experimental.nvidia.gpu
34+
torch-spec: '--pre torch --index-url https://download.pytorch.org/whl/nightly/cu121'
35+
gpu-arch-type: "cuda"
36+
gpu-arch-version: "12.1"
37+
38+
uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
39+
with:
40+
timeout: 300
41+
runner: ${{ matrix.runs-on }}
42+
gpu-arch-type: ${{ matrix.gpu-arch-type }}
43+
gpu-arch-version: ${{ matrix.gpu-arch-version }}
44+
submodules: recursive
45+
script: |
46+
conda create -n venv python=3.9 -y
47+
conda activate venv
48+
export PATH=/opt/rh/devtoolset-10/root/usr/bin/:$PATH
49+
python -m pip install --upgrade pip
50+
pip install ${{ matrix.torch-spec }}
51+
pip install -r dev-requirements.txt
52+
pip install .
53+
cd torchao/prototype/float8nocompile
54+
pytest kernels/ --verbose -s
55+
pytest test/train_test.py --verbose -s

‎.github/workflows/nightly_smoke_test.yml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -21,9 +21,9 @@ jobs:
2121
include:
2222
- name: CUDA Nightly
2323
runs-on: linux.g5.12xlarge.nvidia.gpu
24-
torch-spec: '--pre torch --index-url https://download.pytorch.org/whl/nightly/cu121'
24+
torch-spec: '--pre torch==2.7.0.dev20250122 --index-url https://download.pytorch.org/whl/nightly/cu124'
2525
gpu-arch-type: "cuda"
26-
gpu-arch-version: "12.1"
26+
gpu-arch-version: "12.4"
2727

2828
permissions:
2929
id-token: write

‎.github/workflows/regression_test.yml

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -25,12 +25,12 @@ jobs:
2525
include:
2626
- name: CUDA Nightly
2727
runs-on: linux.g5.12xlarge.nvidia.gpu
28-
torch-spec: '--pre torch --index-url https://download.pytorch.org/whl/nightly/cu124'
28+
torch-spec: '--pre torch==2.7.0.dev20250122 --index-url https://download.pytorch.org/whl/nightly/cu124'
2929
gpu-arch-type: "cuda"
3030
gpu-arch-version: "12.4"
3131
- name: CPU Nightly
3232
runs-on: linux.4xlarge
33-
torch-spec: '--pre torch --index-url https://download.pytorch.org/whl/nightly/cpu'
33+
torch-spec: '--pre torch==2.7.0.dev20250122 --index-url https://download.pytorch.org/whl/nightly/cpu'
3434
gpu-arch-type: "cpu"
3535
gpu-arch-version: ""
3636

‎.gitignore

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -262,7 +262,7 @@ docs/dev
262262
docs/build
263263
docs/source/tutorials/*
264264
docs/source/gen_modules/*
265-
docs/source/sg_execution_times
265+
docs/source/sg_execution_times.rst
266266

267267
# LevelDB files
268268
*.sst

‎docs/source/_templates/layout.html

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@
22

33
{% block sidebartitle %}
44
<div class="version">
5-
<a href='tba'>{{ version }} &#x25BC</a>
5+
<a href='index.html'>{{ version }} &#x25BC</a>
66
</div>
77
{% include "searchbox.html" %}
88
{% endblock %}
@@ -22,7 +22,7 @@
2222
// to point to the torchao repo.
2323
var overwrite = function (_) {
2424
if ($(this).length > 0) {
25-
$(this)[0].href = "https://github.com/pytorch-labs/ao"
25+
$(this)[0].href = "https://github.com/pytorch/ao"
2626
}
2727
}
2828
// PC

‎docs/source/api_ref_dtypes.rst

Lines changed: 28 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -6,19 +6,42 @@ torchao.dtypes
66

77
.. currentmodule:: torchao.dtypes
88

9+
Layouts and Tensor Subclasses
10+
-----------------------------
11+
.. autosummary::
12+
:toctree: generated/
13+
:nosignatures:
14+
15+
NF4Tensor
16+
AffineQuantizedTensor
17+
Layout
18+
PlainLayout
19+
SemiSparseLayout
20+
TensorCoreTiledLayout
21+
Float8Layout
22+
FloatxTensor
23+
FloatxTensorCoreLayout
24+
MarlinSparseLayout
25+
BlockSparseLayout
26+
UintxLayout
27+
MarlinQQQTensor
28+
MarlinQQQLayout
29+
Int4CPULayout
30+
CutlassInt4PackedLayout
31+
32+
Quantization techniques
33+
-----------------------
934
.. autosummary::
1035
:toctree: generated/
1136
:nosignatures:
1237

13-
to_nf4
1438
to_affine_quantized_intx
1539
to_affine_quantized_intx_static
40+
to_affine_quantized_fpx
1641
to_affine_quantized_floatx
1742
to_affine_quantized_floatx_static
18-
to_affine_quantized_fpx
19-
NF4Tensor
20-
AffineQuantizedTensor
21-
43+
to_marlinqqq_quantized_intx
44+
to_nf4
2245
..
2346
_NF4Tensor - add after fixing torchao/dtypes/nf4tensor.py:docstring
2447
of torchao.dtypes.nf4tensor.NF4Tensor.dequantize_scalers:6:Unexpected indentation.

‎docs/source/api_ref_sparsity.rst

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -12,7 +12,7 @@ torchao.sparsity
1212

1313
WandaSparsifier
1414
PerChannelNormObserver
15-
apply_sparse_semi_structured
1615
apply_fake_sparsity
17-
18-
16+
sparsify_
17+
semi_sparse_weight
18+
int8_dynamic_activation_int8_semi_sparse_weight

0 commit comments

Comments
 (0)