Commits · 99a1c74425e0e785bada2c1a19dfb79a343e5310 · OpenDAS / TransformerEngine

04 Feb, 2026 2 commits
- Fix undefined use_int8 error · 99a1c744
  wenjh authored Feb 04, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  99a1c744
- Remove dump code of tensorwise_int8_bgrad_kernel · 2bb532fb
  wenjh authored Feb 04, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  2bb532fb
30 Jan, 2026 1 commit
- Fix out-of-bounds issues for types struct in common/common.h · d2c77acc
  wenjh authored Jan 30, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  d2c77acc
23 Jan, 2026 4 commits

Fix issues related to L1cpp tests · 284d3f6f

maxiao3 authored Jan 23, 2026



1,not find nvte_dgelu
2,fsdp_group is not none
3,CPUOffloadEnabled change to cpp_offload_v1
Signed-off-by: maxiao3 <maxiao3@sugon.com>

See merge request dcutoolkit/deeplearing/TransformerEngine!74

284d3f6f

Fix issues related to L0cpp tests · 8fc9d8f1

maxiao3 authored Jan 23, 2026



1,Resolve out-of-bounds issues for types struct
2,Fix TestFusedCastFloat8Vectorwise test case failure
Signed-off-by: maxiao3 <maxiao3@sugon.com>

See merge request dcutoolkit/deeplearing/TransformerEngine!73

8fc9d8f1

[DCU] Remove redundant shared memory in rowwise kernel · 261e476b

zc20020701 authored Jan 23, 2026


Signed-off-by: zhaochao <zhaochao1@sugon.com>

See merge request dcutoolkit/deeplearing/TransformerEngine!72
Co-authored-by: zhaochao <zhaochao1@sugon.com>

261e476b

Refine the constraints while using lightop in gemm.py · 6c9dc19d
wenjh authored Jan 23, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
6c9dc19d

21 Jan, 2026 1 commit

Add NVTE_USE_LIGHTOP env var to control lightop import · 59b49b47

maxiao3 authored Jan 21, 2026


Signed-off-by: maxiao3 <maxiao3@sugon.com>

See merge request dcutoolkit/deeplearing/TransformerEngine!71

59b49b47

12 Jan, 2026 1 commit
- Fix building on nmz · 0fce42f7
  wenjh authored Jan 12, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  0fce42f7
09 Jan, 2026 2 commits

Develop v2.10 · 13123839

dongchl authored Jan 09, 2026



rollback activation offloading implementation

See merge request dcutoolkit/deeplearing/TransformerEngine!70
Co-authored-by: dongcl <791582849@qq.com>

13123839

Fix swizzle, swap_first_dims and RMSNorm issues · e6f2caf5
wenjh authored Jan 09, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
e6f2caf5

08 Jan, 2026 1 commit

Fix tests of L0 test_numeric and L1 test_fusible_ops · 953b6d68

wenjh authored Jan 08, 2026


Signed-off-by: wenjh <wenjh@sugon.com>

See merge request dcutoolkit/deeplearing/TransformerEngine!67

953b6d68

07 Jan, 2026 2 commits
- Add nmz support · dc86f372
  wenjh authored Jan 07, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  dc86f372
- Rename package for tefl · 08be824c
  wenjh authored Jan 07, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  08be824c
05 Jan, 2026 3 commits
- Rename tefl hygon backend · bdf3d931
  wenjh authored Jan 05, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  bdf3d931
- Make hygon backend installable · 40816696
  wenjh authored Jan 05, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  40816696
- Add hygon backend for TE-FL · 73d959a4
  wenjh authored Jan 05, 2026
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  73d959a4
31 Dec, 2025 5 commits
- Using pip install to build · b0871127
  wenjh authored Dec 31, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  b0871127
- Add bias fwd/bwd at group gemm · cb2fe806
  wenjh authored Dec 31, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  cb2fe806
- Fix new gemm · f796eb80
  wenjh authored Dec 31, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  f796eb80
- Enable test_attention · 04afba37
  wenjh authored Dec 31, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  04afba37
- Update version to das1.8 · ea5cc27a
  wenjh authored Dec 31, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  ea5cc27a
27 Dec, 2025 1 commit
- Merge nv_main(2.10) to main · c1a1c04e
  wenjh authored Dec 27, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  c1a1c04e
15 Dec, 2025 2 commits

Fix blaslt group gemm crush · e698a0a7
wenjh authored Dec 15, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
e698a0a7

Merge branch 'TE_ta_develop2.9' into 'develop_v2.9' · 4086a4cc

wenjh authored Dec 15, 2025

[DCU]Fix compilation unable to find nvte-extract_ded_and_offset

See merge request dcutoolkit/deeplearing/TransformerEngine!64

4086a4cc

10 Dec, 2025 2 commits
- [DCU]fix mip and compile issues · 121d9224
  tabuchixiangcai3 authored Dec 10, 2025
```
Signed-off-by: Tangao <2205747538@qq.com>
```
  121d9224
- [DCU]Fix compilation unable to find nvte-extract_ded_and_offset · ba058648
  tabuchixiangcai3 authored Dec 10, 2025
```
Signed-off-by: Tangao <2205747538@qq.com>
```
  ba058648
04 Dec, 2025 2 commits
- Merge branch 'develop_v2.9' into 'develop_v2.9' · 177291ac
  wenjh authored Dec 04, 2025
```
grouped_linear supports activation offloading

See merge request dcutoolkit/deeplearing/TransformerEngine!62
```
  177291ac
- grouped_linear supports activation offloading · db0ad945
  dongcl authored Dec 04, 2025
  
  db0ad945
03 Dec, 2025 5 commits
- Make release_v2.9 compile pass · 99e60246
  wenjh authored Dec 03, 2025
  
  99e60246
- Fix build error · cbb14a5f
  wenjh authored Dec 03, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  cbb14a5f
- Fix build error · b3dcfc28
  wenjh authored Dec 03, 2025
  
  b3dcfc28
- Update build and install command · 1e3c6a25
  wenjh authored Dec 03, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  1e3c6a25
- Merge nv main up to v2.10.0.dev0 · 063ef88d
  wenjh authored Dec 03, 2025
```
Signed-off-by: wenjh <wenjh@sugon.com>
```
  063ef88d
26 Nov, 2025 2 commits

Merge branch 'develop_v2.8' into 'main' · 91670b05

wenjh authored Nov 26, 2025

[DCU] Skip some tests in test_sanity.py

See merge request dcutoolkit/deeplearing/TransformerEngine!61

91670b05

Merge branch 'fix_develop2.8_zc' into 'develop_v2.8' · 3a040217
wenjh authored Nov 26, 2025
```
[DCU]Fix some bugs

See merge request dcutoolkit/deeplearing/TransformerEngine!56
```
3a040217

15 Nov, 2025 1 commit
- Updated VERSION to 2.11.0.dev0 · 66aed3ae
  Przemek Tredak authored Nov 14, 2025
```
Signed-off-by: Przemek Tredak <ptredak@nvidia.com>
```
  66aed3ae
14 Nov, 2025 3 commits

[JAX] Use TE quant if TE fused act is disabled (#2374) · 389a6ba4

jberchtold-nvidia authored Nov 14, 2025



* Use TE quant if TE fused act is disabled
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Keep existing precision
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

---------
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

389a6ba4

[PyTorch] Activation offloading refactor (#1762) · c5257605

Paweł Gadziński authored Nov 14, 2025



* init
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* offloading
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* all types
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* typo
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* init
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* api change
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* code drop
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* refactor
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* tests
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* code drop
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* example
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* cpu offload + debug warning
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* test
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* change empty_like implementation to use make_like
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* main_grad fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* manual synchornization
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* old path
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* remove example
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* api changes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* reverted grouped linear
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* make odl code path work for modules
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* attention old code path
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* legacy tests
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* legacy tests
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* updated code path
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Update transformer_engine/pytorch/tensor/quantized_tensor.py
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Signed-off-by: Paweł Gadziński <62263673+pggPL@users.noreply.github.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* nvfp4 support
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fixes
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* Update tests/pytorch/test_cpu_offloading.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Signed-off-by: Paweł Gadziński <62263673+pggPL@users.noreply.github.com>

* small fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* docs change
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

---------
Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>
Signed-off-by: Paweł Gadziński <62263673+pggPL@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: root <root@ptyche0312.ptyche.clusters.nvidia.com>
Co-authored-by: Tim Moon <4406448+timmoon10@users.noreply.github.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>

c5257605

[JAX] Improve support and testing for direct recipe usage without autocast contexts (#2366) · a0754757

jberchtold-nvidia authored Nov 14, 2025



* Refactor to avoid storing a global quantization config so direct recipe passing works as intended
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* fix use_split_accumulator for current scaling recipe
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* fix tests that pass direct recipe and were missing quantize meta set
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Revert "fix use_split_accumulator for current scaling recipe"

This reverts commit a74ab7df812ec0a069b1bdd208debb93ec25a900.
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* fix ci failures
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* Fix amax_history post_init
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Signed-off-by: jberchtold-nvidia <158520091+jberchtold-nvidia@users.noreply.github.com>

* Update transformer_engine/jax/quantize/quantizer.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Signed-off-by: jberchtold-nvidia <158520091+jberchtold-nvidia@users.noreply.github.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci



* fix ci failures
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* fix ci issue
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* address comments
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

* make recipe assertion classes in test_recipe_characteristics not inherit from unittest.TestCase
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>

---------
Signed-off-by: Jeremy Berchtold <jberchtold@nvidia.com>
Signed-off-by: jberchtold-nvidia <158520091+jberchtold-nvidia@users.noreply.github.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

a0754757