fzyzcjy
3b24d8981b
Report per-token weight-version spans in generation meta info ( #35926 )
2026-08-24 20:18:52 +08:00
fzyzcjy
981dfa2b83
Make the scheduler track the published weight version ( #35925 )
2026-08-24 20:17:50 +08:00
fzyzcjy
02a3dca738
Extract _make_abort_req from the scheduler's abort paths ( #35924 )
2026-08-24 20:16:38 +08:00
fzyzcjy
a37fdae562
Extract collect_inflight_reqs from abort_request for reusing ( #35923 )
2026-08-24 20:15:09 +08:00
Mick
9856b58de4
[diffusion] feat: support loading serialized fp8 clip image encoders ( #36056 )
2026-08-24 19:59:35 +08:00
Mick
bfeae4e79a
[diffusion] feat: support loading serialized convrot w4a4 checkpoints ( #36039 )
2026-08-24 19:13:35 +08:00
guzekai01
716a6bf10c
feat(humming): support native W4AFP8 checkpoint schemas ( #32033 )
2026-08-24 18:59:27 +08:00
guzekai01
21258b7a35
feat(humming): FP8 DeepEP dispatch for humming MoE backend ( #31429 )
2026-08-24 18:59:07 +08:00
Xiaoyu Zhang
46b92b22e2
[diffusion] Accelerate LingBot Video RMSNorm in quality=high ( #35969 )
2026-08-24 18:02:00 +08:00
092d85eb87
[Feature] Add MiniCPM-SALA support ( #30360 )
...
Co-authored-by: Alex Nails <alex.nails@radixark.ai >
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com >
2026-08-24 02:25:16 -07:00
SuperSong
d251fa2453
perf(unified-memory): batch lazy-compaction mapping lookup ( #34066 )
2026-08-24 02:14:45 -07:00
Mick
51b27f747a
[diffusion] feat: support loading pruned minimax h3 components natively ( #36070 )
2026-08-24 16:39:37 +08:00
Mick
adc09a1f63
[diffusion] feat: support loading self-describing quanto int8 encoders ( #36052 )
2026-08-24 16:37:09 +08:00
Mick
5081ad5d4e
[diffusion] feat: add per-component quantization overrides ( #36084 )
2026-08-24 16:35:48 +08:00
jiayisunx
5c4622341f
[GDN] remove XPU path of causal_conv1d_fn and causal_conv1d_update ( #35775 )
2026-08-24 15:59:23 +08:00
Alex Nails and Claude Opus 5
2070927e00
[Triton] Bound the sliding-window extend-attention KV loop: -86.6% on SWA layers, -9.4% prefill GPU, bit-identical ( #34462 )
...
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com >
2026-08-24 00:58:52 -07:00
7bbd0ddeb5
[AMD] Quark shared-experts gate: recognise a trailing MTP layer ( #36124 )
...
Co-authored-by: Cursor <cursoragent@cursor.com >
Co-authored-by: Thomas Wang <thomawan@amd.com >
2026-08-24 00:53:52 -07:00
Alex Nails and Claude Opus 5
666b08b4a5
[ROCm] Extend the gfx950 extend-attention tile to head_dim <= 128: -43% kernel, -14% TTFT, bit-identical ( #34461 )
...
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com >
2026-08-24 00:29:39 -07:00
McZyWu
c439e77872
[bugfix] [NPU] fix transpose batch matmul K*B exceed 65536. ( #34715 )
2026-08-24 15:22:59 +08:00
Mick and Claude Fable 5
c8e1ddc707
[diffusion] feat: cache LoRA-merged weights in files the page cache can hold ( #36062 )
...
Co-authored-by: Claude Fable 5 <noreply@anthropic.com >
2026-08-24 14:18:55 +08:00
Xiaoyu Zhang
9866fe910b
[diffusion] Speed up LingBot high-quality VAE decode ( #36024 )
2026-08-24 14:13:03 +08:00
Xiaoyu Zhang
cc74aba330
[diffusion] Honor XDG cache for model overlays ( #36019 )
2026-08-24 14:11:26 +08:00
ming_wang
97b176e64c
Support streaming session on NPU ( #32597 )
2026-08-24 14:07:44 +08:00
Shangming Cai
852b04b358
fix(xpu): read enable_deterministic_inference from the config bag ( #36149 )
2026-08-24 13:56:41 +08:00
Xiaoyu Zhang
6d40b8aebf
[diffusion] Fix Hunyuan QKV pack indexing at production video shapes ( #36009 )
2026-08-24 13:39:26 +08:00
Xiaoyu Zhang
b43931e878
[diffusion] Refresh quality and BCG benchmark skills ( #36016 )
...
Signed-off-by: BBuf <1182563586@qq.com >
2026-08-24 13:37:45 +08:00
Yuan Luo and luoyuan.luo
77940dec80
[MoE] Gather the cutlass MoE activation and its scales in one launch ( #34915 )
...
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com >
2026-08-24 13:37:14 +08:00
Mick
3fe18f13cd
[diffusion] feat: add plain component weight overrides ( #36086 )
2026-08-24 13:33:23 +08:00
Mick
8df3b9eff9
[diffusion] feat: support loading mixed w4a8 text encoders ( #36037 )
2026-08-24 13:31:27 +08:00
Xiaoyu Zhang
344613c159
[diffusion] Default Hunyuan VAE to tiled decode ( #36012 )
2026-08-24 13:18:06 +08:00
vikram singh shekhawat
f98b60de80
fix(xpu): enable compressed-tensors FP8 W8A8 on XPU (RedHatAI FP8-dynamic models) ( #33057 )
2026-08-24 12:53:48 +08:00
Lianmin Zheng
b498efce52
chore: move cuda_vmm_utils.py under srt/utils/ ( #36053 )
2026-08-23 21:47:29 -07:00
gaopengff and Ma Mingfei
56834422a1
[Intel XPU] Add xpu pass for biased_topk and hash_topk ( #33323 )
...
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com >
2026-08-24 12:18:22 +08:00
Chunyuan WU
1daa94a069
[CPU] Fix NUMA/core binding for DP ranks ( #32856 )
2026-08-24 12:15:38 +08:00
Aurick Qiao and Yuwei An
3e30649064
[Fix] Harden FlashAttention CUDA graph metadata bounds ( #35454 )
...
Co-authored-by: Yuwei An <ayw.sirius19@gmail.com >
2026-08-23 21:11:46 -07:00
Xia Weiwen
5b5b29d4e2
[XPU] Use a fused GDN kernel from sgl-kernel for Qwen3.5 ( #33354 )
2026-08-24 12:07:50 +08:00
Xiaoyu Zhang
8dcfb3b5e7
[diffusion] Fuse LongCat-Image QKNorm and interleaved RoPE ( #35995 )
2026-08-24 12:07:26 +08:00
Xiaoyu Zhang
09592f5889
[diffusion] Keep LongLive2 components resident on large GPUs ( #35993 )
2026-08-24 12:06:52 +08:00
Xuan Liao and Ma Mingfei
acba8921bf
[XPU] Support softmax_lse in sgl_kernel::fwd API ( #33840 )
...
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com >
2026-08-24 12:05:15 +08:00
Mick
7a7b655ddf
[quantization] share bounded post-load device staging ( #35180 )
2026-08-24 11:48:30 +08:00
Mick
5ce700aee8
[diffusion] feat: infer LoRA alpha from safetensors metadata ( #36082 )
2026-08-24 11:46:18 +08:00
Mick
230c052ebc
[diffusion] chore: reuse srt AutoRound for quantized DiTs ( #36068 )
2026-08-24 11:21:44 +08:00
Mick and Claude Opus 5
f294d51a71
[diffusion] fix: fix a refit key error on mapped weights, and stop claiming strides the reload discards ( #35832 )
...
Co-authored-by: Claude Opus 5 <noreply@anthropic.com >
2026-08-24 11:11:35 +08:00
Alison Shao
0c1e9bda57
[OpenAI] Drop empty assistant turns for mistral_common tokenizers ( #35915 )
...
mistral_common rejects an assistant turn carrying neither content nor tool calls, while other chat templates ignore it, so an OpenAI-compatible request that works elsewhere failed on Mistral models. Drop those turns before templating; turns with tool_calls, multimodal content, or real text are untouched, and a trailing assistant message is already consumed upstream as the continue_final_message prefix.
2026-08-23 20:07:17 -07:00
Mick
6ca872a11f
[diffusion] chore: fetch metadata beside nested lora weights ( #36057 )
2026-08-24 11:02:23 +08:00
Mick
1a368eca1c
[diffusion] optimization: reuse minimax h3 prompt refinement across outputs ( #36027 )
2026-08-24 11:01:21 +08:00
Xun Sun and siyu
a90d770c40
[Weight Cache] Support static DP/EP layouts ( #33684 )
...
Co-authored-by: siyu <liusy58@linux.alibaba.com >
2026-08-23 19:57:52 -07:00
YangKai0616 and gemini-code-assist[bot]
fbdec2855a
[XPU] Support INT4 dense linear (AWQ/GPTQ) for XPU ( #30236 )
...
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2026-08-24 10:41:59 +08:00
Mick
f6fff25756
[diffusion] feat: support vae weight-file overrides ( #36085 )
2026-08-24 10:33:14 +08:00
Mick
2d84de5e69
[diffusion] feat: support loading serialized comfy w4a8 checkpoints ( #36036 )
2026-08-24 10:32:21 +08:00
fzyzcjy