4 people 13d593b6cf dsv4.1: compression, KV I/O, and metadata kernels (#39652) ...

Co-authored-by: BBuf <1182563586@qq.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: DarkSharpness <76582120+DarkSharpness@users.noreply.github.com>

2026-09-16 13:54:07 -07:00

..

attention

dsv4.1: compression, KV I/O, and metadata kernels (#39652)

2026-09-16 13:54:07 -07:00

communication

[Feature] Support BF16 and batch-invariant inference with DeepEP v2 (#38160)

2026-09-15 14:30:50 -07:00

cuda_graph

Reuse live CUDA graph executables during dedup registration (#39176)

2026-09-12 11:44:17 -07:00

diffusion

[diffusion] model: support VDN-H3 with a hybrid_window_attn_h3 backend (#37903)

2026-09-12 11:36:32 +08:00

disaggregation

[PD] Add optional KV transfer checksums (#39500)

2026-09-16 23:53:30 +08:00

embeddings

[Qwen3.8] Enable NVIDIA NVFP4 on DGX Spark with file-backed PLE and PDL router fix (#39126)

2026-09-13 16:23:41 +08:00

hyperconnection

support qwen 3.8 flash next (#37500)

2026-09-08 13:56:21 -07:00

jit

support qwen 3.8 flash next (#37500)

2026-09-08 13:56:21 -07:00

layernorm

dsv4.1: standalone kernels and Python wrappers (#39646)

2026-09-15 22:31:18 -07:00

memory

[PD] Add optional KV transfer checksums (#39500)

2026-09-16 23:53:30 +08:00

moe

[Feature] Support BF16 and batch-invariant inference with DeepEP v2 (#38160)

2026-09-15 14:30:50 -07:00

ops/attention

Store mamba prefix-cache checkpoints at the configured SSM state dtype (#34820)

2026-09-09 15:25:57 +08:00

qsa

[Qwen 3.8 Next] Remove unused tokenwise QSA implementation and tests (#38960)

2026-09-12 01:32:01 -07:00

quantization

[AMD] Fix weight checking for AITER-shuffled block FP8 weights (#34330)

2026-09-11 15:49:56 -07:00

speculative

[AMD] Parallelize aiter spec-decode KV index building over token blocks (#37659)

2026-09-09 17:53:53 -07:00