merge main#42

Merged

yukavio merged 2440 commits intoyukavio:sparse_attn_fa3from

josephydu:sparse_attn_fa3_merge

Nov 25, 2025

This pull request is big! We're only showing the most recent 250 commits

Commits on Nov 17, 2025

Commits on Nov 18, 2025

Commits on Nov 19, 2025

Commits on Nov 20, 2025

[Auto Sync] Update base_grammar_backend.py, collector.py (20251116) (#13357)

authored
[GDN] Remove unnecessary contiguous() (#13604)
byjiang1996
authored
[GDN] Remove unnecessary conv state clone (#13603)
byjiang1996
authored
[VLM] Support Piecewise CUDA Graph for Qwen2.5-VL (#13055)

authored
[diffusion] CI: improve diffusion CI (#13562)

mickqian
and
adarshxs
authored
feat: support external custom models (#13429)

authored
[CI fix] Fix image download failures in VLM CI tests (#13613)
BBuf
authored
[NVIDIA] Add fp8 gemm benchmark on blackwell (#13528)
kaixih
authored
[UT] Destroy process group after broadcast to resolve port occupation issues in multi-server tests (#12379)
galeselee
authored
[diffusion] refactor: remove PreprocessorConfig (#13248)
mickqian
authored
[diffusion] refactor: refactor pipeline folders (#13253)
mickqian
authored
Add FP32 dtype support for RoPE - Part2 (#13328)
jinyouzhi
authored
[diffusion] fix: remove multimodal_gen redundant get_bool_env_var func (#13583)

shauntajoesph-ops
and
mickqian
authored
Add support for new aiter version (AR accuracy, is_shuffled PR) (#13554)

1am9trash
and
sogalin
authored
diffusion: improve baseline performance monitor (#13614)
mickqian
authored
[Feature] Introduce JIT Kernel in sglang (with hicache JIT kernel) (#13453)
DarkSharpness
authored
[CI] Align metric units for CI rate limit (#13633)

hnyls2002
and
gemini-code-assist[bot]
authored
[ROCM] Optimized deepseek-r1 fp8 model with + triton_gemm_a8w8 + batch_gemm_a8w8 + fused set_mla_kv_buffer kernel (#13617)

authored
fix bench_speculative bug (#13197)
Lzhang-hub
authored
Revert "[Feature] Introduce JIT Kernel in sglang (with hicache JIT kernel)" (#13644)
merrymercy
authored
[CI] optimize CI workflow info (#13634)
hnyls2002
authored
CI: Kill zombie diffusion processes in CI & minor code style fix on rotary embedding fallback (#13637)
merrymercy
authored
[CI] apply pr-gate for XPU (#13663)
hnyls2002
authored
Add fused_rmsnorm_gated_cpu kernel for CPU to support Qwen3-Next (#11577)
yanbing-j
authored
[10/n] decouple quantization impl from vllm dependency - fix import (#13524)
FlamingoPg
authored
Adding nightly tests as release guard for bot bump workflows (#13655)
dougyster
authored
[DeepseekV3.2] Deepseek fp8 support for MHA path (#12964)
YAMY1234
authored
Fix launch of `Olmo3` (#13666)
vincentzed
authored
[Deepseek V3.2] Change indexer weights_proj to fp32 (#13459)
hlu1
authored
enable csgmv automatically on cuda (#13600)
b8zhong
authored
Add nightly test CI monitor workflow (#13038)
alisonshao
authored
allow loras to be implicitly evicted and loaded based on max_loaded_loras (#11526)
glenliu21
authored
Test reorganization: Move tests to manual/ (#13610)
alisonshao
authored
[Piecewise CUDA Graph] Fix recompile issue for Mixtral and Grok2 (#13667)

authored
Super tiny remove unused MiniMaxM2MLP class (#13659)
fzyzcjy
authored
Update quantization.md with new model resources (#13677)
zhaochenyang20
authored

Commits on Nov 21, 2025

Commits on Nov 22, 2025

Commits on Nov 23, 2025

Commits on Nov 24, 2025

Commits on Nov 25, 2025