Skip to content

Pull requests: pytorch/ao

New pull request New
205 Open 2,587 Closed
205 Open 2,587 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

adding Int8 weight only and dynamic configs to safetensors CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: improvement Use this tag if this PR is an improvement (doesn't fit into any of the other categories)
#3474 opened Dec 9, 2025 by liangel-02 Loading…
1
make eval script also handle performance measurement CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: for developers Use this tag if this PR is mainly developer facing
#3473 opened Dec 9, 2025 by vkuzo Loading…
2
refactor accuracy eval script to be organized by hardware CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: for developers Use this tag if this PR is mainly developer facing
#3472 opened Dec 9, 2025 by vkuzo Loading…
3
use python version agnostic c++ extension for mxfp8_cuda CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: bug fix Use this tag for PRs that fix bugs
#3471 opened Dec 9, 2025 by danielvegamyhre Loading…
3
simplify accuracy eval CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: for developers Use this tag if this PR is mainly developer facing
#3470 opened Dec 9, 2025 by vkuzo Loading…
2
enable smoothquant for int8 static tensor CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: improvement Use this tag if this PR is an improvement (doesn't fit into any of the other categories)
#3468 opened Dec 8, 2025 by jcaip Loading…
5
Support HQQ for XPU/NPU int4 quantization CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: improvement Use this tag if this PR is an improvement (doesn't fit into any of the other categories)
#3465 opened Dec 8, 2025 by namgyu-youn Loading…
7
support int8 scaled embedding bag pattern match CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3463 opened Dec 8, 2025 by shiyang-weng • Draft
1
Fix type annotation of _choose_qparams_affine/gguf(): input = None CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3461 opened Dec 6, 2025 by Freed-Wu Loading…
2
Fix #3458, correct min_val and max_val CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3460 opened Dec 6, 2025 by Freed-Wu Loading…
1
Fix bug of _quantize_affine_no_dtype_cast() when zero_point is None CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3459 opened Dec 6, 2025 by Freed-Wu Loading…
1
Add nightly regression tests for CUDA 12.8 and 12.9 on H100/A100 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: not user facing Use this tag if you don't want this PR to show up in release notes
#3448 opened Dec 5, 2025 by jainapurva • Draft
3
Align _choose_qparams_affine with _choose_scale_float8 behavior CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3447 opened Dec 5, 2025 by AryanBagade Loading…
10
Remove unused exception parameter from multifeed/recommendation_platform/corpus/backfill/BackfillMain.cpp CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. fb-exported meta-exported module: rocm
#3445 opened Dec 5, 2025 by r-barnes Loading…
2
[bc-breaking] update default int8 tensor version to v2 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: bc-breaking Use this tag if this PR breaks backward compatibility
#3441 opened Dec 5, 2025 by jcaip Loading…
2
use the float4 dtype in mxfp4 and nvfp4 tensors CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: not user facing Use this tag if you don't want this PR to show up in release notes
#3440 opened Dec 4, 2025 by vkuzo Loading…
2
Update shapes in microbenchmarks to reflect more realistic data ciflow/benchmark CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: not user facing Use this tag if you don't want this PR to show up in release notes
#3438 opened Dec 4, 2025 by jainapurva Loading…
1
Add support for MXFP8 All gather CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: improvement Use this tag if this PR is an improvement (doesn't fit into any of the other categories)
#3435 opened Dec 4, 2025 by avizon-aws Loading…
28
[reland][xpu][test] Port 2 test/prototype/test_{parq, quantized_training} UT files to intel XPU ci-no-td ciflow/xpu label used to trigger xpu CI jobs CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: for developers Use this tag if this PR is mainly developer facing xpu Intel XPU related features
#3434 opened Dec 4, 2025 by zxd1997066 Loading…
2
[WIP] Merge all model eval scripts together CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3430 opened Dec 3, 2025 by jainapurva • Draft
1
Add pruning-aware training in torchao.prototype.pat CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: new feature Use this tag if this PR adds a new feature
#3429 opened Dec 3, 2025 by lisjin Loading…
3
not for land: vllm bench for mxfp8 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3426 opened Dec 3, 2025 by vkuzo Loading…
2
[xpu][test] Port test_float8_utils.py and test_hqq_affine.py UT files to intel XPU CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: for developers Use this tag if this PR is mainly developer facing xpu Intel XPU related features
#3424 opened Dec 3, 2025 by zxd1997066 Loading…
3
[xpu][test] Port 2 test/prototype/test_{nvfp4_tensor, spinquant} UT files to intel XPU CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3423 opened Dec 3, 2025 by zxd1997066 Loading…
2
Static smooth quant CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#3422 opened Dec 3, 2025 by cyxlily Loading…
1
ProTip! Add no:assignee to see everything that’s not assigned.