All issues

Search results

[PyTorch][Attention] THD P2P context-parallel regression when padded cu_seqlens are value-equal but not object-identical

attention

Status: Open.

#3487 In NVIDIA/TransformerEngine;

· cuichenx opened on Sep 5, 2026

[Bug] Backend selection picks FA3 for training with head_dim_qk=192 / v_head_dim=128, but FA3 backward cannot run it

attention

Status: Open.

#3481 In NVIDIA/TransformerEngine;

· yuweih205 opened on Sep 4, 2026

[Bug] Illegal memory access in grouped MXFP8 quantize on GB200 (sm_100) introduced by #3337; TMA store completion handoff missing in group_quantize_mxfp8.cuh

Status: Open.

#3474 In NVIDIA/TransformerEngine;

· janbernloehr opened on Sep 3, 2026

Tied weights do not accumulate with delayed wgrad

Status: Open.

#3437 In NVIDIA/TransformerEngine;

· wujingyue opened on Aug 28, 2026

NCCL EP hard link makes import transformer_engine require libcuda.so.1

Status: Open.

#3381 In NVIDIA/TransformerEngine;

· cuichenx opened on Aug 13, 2026

[Bug] Triton mask-map moe_permute ignores num_out_tokens

Status: Open.

#3346 In NVIDIA/TransformerEngine;

· truong-v opened on Aug 11, 2026

[Bug] Deferred init issue in te.pytorch.ops

bugSomething isn't workingSomething isn't working

Status: Open.

#3322 In NVIDIA/TransformerEngine;

· yaox12 opened on Aug 6, 2026

[Bug] CUDA graph backward replay illegal memory access after validation with mixed NVFP4/MXFP8 Mamba

Status: Open.

#3316 In NVIDIA/TransformerEngine;

· Phlip79 opened on Aug 4, 2026

thd + attention_dropout routes to the composite cuDNN SDPA engine and costs 5x on Blackwell

attention

Status: Open.

#3312 In NVIDIA/TransformerEngine;

· bzantium opened on Aug 4, 2026

[Build] Publish PyTorch 2.11 CUDA 13 aarch64 wheels for GB200

Status: Open.

#3310 In NVIDIA/TransformerEngine;

· bvolpato opened on Aug 4, 2026

[Build] Restore CUDA 13 Linux aarch64 core wheels

Status: Open.

#3309 In NVIDIA/TransformerEngine;

· bvolpato opened on Aug 4, 2026

[Bug] Transformer Engine SM120 (Blackwell) Compatibility Issue

bugSomething isn't workingSomething isn't working

Status: Open.

#3299 In NVIDIA/TransformerEngine;

· lmcl90 opened on Aug 3, 2026