.. include:: /content/common.rsts

.. |ge|      replace::  :html:`≥`

Release Notes |ndash| Release 2.11
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!


Key Features and Enhancements
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@

- [PyTorch] Enabled the reference Current Scaling recipe for FP8 training. (`#2368 <https://github.com/NVIDIA/TransformerEngine/pull/2368>`_)
- [PyTorch] Improved Random Hadamard Transform (RHT) device tensor caching to reduce memory allocations and improve performance for NVFP4 quantization. (`#2395 <https://github.com/NVIDIA/TransformerEngine/pull/2395>`_)
- [PyTorch] Implemented selective activation checkpointing for ``LayerNormMLP`` module. (`#2311 <https://github.com/NVIDIA/TransformerEngine/pull/2311>`_)
- [C, PyTorch, JAX] Improved performance of MXFP8 quantization. (`#2062 <https://github.com/NVIDIA/TransformerEngine/pull/2062>`_)
- [C, PyTorch] Improved performance of NVFP4 quantization. (`#2351 <https://github.com/NVIDIA/TransformerEngine/pull/2351>`_)
- [PyTorch] Improved FSDP2 all-gather performance and added support for ``FusedAdam`` optimizer with FSDP2. (`#2370 <https://github.com/NVIDIA/TransformerEngine/pull/2370>`_)
- [PyTorch] Extended debug tools to support GroupedLinear layers. (`#1953 <https://github.com/NVIDIA/TransformerEngine/pull/1953>`_)
- [JAX] Added Triton kernel bindings for JAX, enabling custom Triton kernels in JAX workflows. (`#2437 <https://github.com/NVIDIA/TransformerEngine/pull/2437>`_)
- [C] Introduced experimental ``NVTEGroupedTensor`` class and helper functions. (`#2388 <https://github.com/NVIDIA/TransformerEngine/pull/2388>`_)
- [C, PyTorch, JAX] Added FP8 support for primary weights in MXFP8 format with partial casting and amax calculations. (`#2055 <https://github.com/NVIDIA/TransformerEngine/pull/2055>`_)
- [JAX] Added support for context parallelism (CP) for THD format and sliding window attention (SWA) using all-gather (AG), striped load balancing with stripe size greater than 1. (`#2379 <https://github.com/NVIDIA/TransformerEngine/pull/2379>`_)
- [JAX] Implemented JAX primitives for token permutation operations on single GPU for mixture-of-experts routing. (`#2473 <https://github.com/NVIDIA/TransformerEngine/pull/2473>`_)
- [PyTorch] Added THD format support for ``max_logit`` clipping and ``MuonClip`` gradient clipping operations. (`#2480 <https://github.com/NVIDIA/TransformerEngine/pull/2480>`_)


Fixed Issues
@@@@@@@@@@@@

- [PyTorch] Fixed a numerical issue when noncontiguous tensor was passed to cross_entropy backward pass. (`#2402 <https://github.com/NVIDIA/TransformerEngine/pull/2402>`_)
- [PyTorch] Fixed CUDA graph execution order for backward weight gradient computation when using chunked layers. (`#2376 <https://github.com/NVIDIA/TransformerEngine/pull/2376>`_)
- [C] Fixed runtime library loading logic to properly handle missing dependencies and load order. (`#2297 <https://github.com/NVIDIA/TransformerEngine/pull/2297>`_)
- [Jax] Removed use of scan loop as the default for ring attention due for improved performance. (`#2503 <https://github.com/NVIDIA/TransformerEngine/pull/2503>`_).


Known Issues in This Release
@@@@@@@@@@@@@@@@@@@@@@@@@@@@

There are no known issues in this release.


Breaking Changes in This Release
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@


There are no breaking changes in this release.


. _DeprecatedFeatures:

Deprecated Features
@@@@@@@@@@@@@@@@@@@

There are no deprecated features in this release.