weireweireandweireweire 55c1963df4 Remove unused draft-extend CUDA graph top-k (#31430) ...

Co-authored-by: weireweire <20922698+weireweire@users.noreply.github.com>

2026-07-30 17:45:01 -07:00

..

arg_groups

Integrate pplx a2a backend (#30756)

2026-07-30 15:33:19 -07:00

batch_invariant_ops

[NPU]adaptation to support deterministic inference (#21197)

2026-06-09 09:22:09 +08:00

batch_overlap

Integrate pplx a2a backend (#30756)

2026-07-30 15:33:19 -07:00

checkpoint_engine

…

compilation

fix: fix vlm cuda graph shape stability (#30868)

2026-07-19 22:35:51 +08:00

configs

[Qwen3.5][MTP] Support FlashInfer CuTe DSL for online NVFP4 draft MoE (#31382)

2026-07-30 17:19:48 -07:00

connector

[connector] Add Azure Blob Storage connector (az:// and *.blob.core.windows.net URLs) (#23995)

2026-05-11 14:20:22 -07:00

constrained

Use native batched llguidance mask generation (#32412)

2026-07-25 16:36:32 -07:00

debug_utils

[Fix] Update PR #25015 revert for fused topk=1 draft postprocess (#31519)

2026-07-16 21:51:16 -07:00

disaggregation

[BugFix][EPD] Early-release mooncake GPU embeddings; fix gpu_id via scheduler.ps (#31591)

2026-07-30 17:36:55 +08:00

distributed

Disable extra NCCL CUDA event synchronization with symm mem (#27089)

2026-07-29 22:39:36 +08:00

dllm

[DLLM] vectorized joint/low-confidence decoding and skip redundant attn init (#21094)

2026-07-30 09:13:14 +03:00

elastic_ep

[Elastic EP] Fix recovery lifecycle and add manual coverage (#31744)

2026-07-25 14:32:47 +08:00

entrypoints

embedding: centralize capabilities and complete OpenAI compatibility (#32481)

2026-07-30 10:28:52 +08:00

eplb

[2/N] elastic-ep: Enable EPLB after scale-up (#30553)

2026-07-30 01:06:56 +08:00

function_call

Fix LFM 2 tool parser. (#27614)

2026-07-30 10:15:08 +00:00

grpc

…

hardware_backend

【NPU】fix decode MTP + eagle shape error (#32791)

2026-07-30 21:34:16 +08:00

kv_canary

Radix Cache Split: Spin off TreeCore (#29901)

2026-07-25 14:31:59 -07:00

layers

[Qwen3.5][MTP] Support FlashInfer CuTe DSL for online NVFP4 draft MoE (#31382)

2026-07-30 17:19:48 -07:00

lora

[LoRA] Fix Marlin MoE kernel import (#32884)

2026-07-30 15:17:29 +08:00

managers

[PD] Handle abort requests in PP mode (#32797)

2026-07-30 14:39:44 +08:00

mem_cache

[Kernel] Fuse KV-cache writes for asymmetric K/V (head_dim != v_head_dim) (#32813)

2026-07-30 00:26:10 -07:00

model_executor

[Qwen3.5][MTP] Support FlashInfer CuTe DSL for online NVFP4 draft MoE (#31382)

2026-07-30 17:19:48 -07:00

model_loader

model: support EmbeddingGemma (#32375)

2026-07-27 10:40:47 +08:00

models

Integrate pplx a2a backend (#30756)

2026-07-30 15:33:19 -07:00

multimodal

vlm: reject moss vision metadata mismatches (#31957)

2026-07-29 07:09:54 +08:00

multiplex

Revert RuntimeContext config-namespace reads/roles (#31813–#31817) (#32100)

2026-07-22 11:52:41 -07:00

observability

fix: prevent ReqTimeStats from being dropped during IPC serialization (#31339)

2026-07-30 14:42:38 +08:00

parser

Flush dropped reasoning at stream end when stream_reasoning=False (#32225)

2026-07-31 08:25:54 +08:00

platforms

[Intel GPU] Add XPU Platform support (#31949)

2026-07-24 12:49:29 +08:00

plugins

…

ray

[FEAT] Support fast engine recovery through weight cache (#27139)

2026-07-25 14:31:21 +08:00

sampling

[Refactor] Move sampling tokenizer validation helper (#32694)

2026-07-28 16:48:03 -07:00

session

[HiCache] Merge HiCache event checks to reduce decode overhead (#30511)

2026-07-29 23:48:48 +08:00

speculative

Remove unused draft-extend CUDA graph top-k (#31430)

2026-07-30 17:45:01 -07:00

state_capturer

Revert RuntimeContext config-namespace reads/roles (#31813–#31817) (#32100)

2026-07-22 11:52:41 -07:00

tokenizer

Add Inkling model support (#31681)

2026-07-19 22:57:37 -07:00

utils

[Fix] Count multi-layer draft-extend replays in the fwd-occupancy device timer (#32867)

2026-07-30 00:21:34 -07:00

weight_cache

[FEAT] Support fast engine recovery through weight cache (#27139)

2026-07-25 14:31:21 +08:00

weight_sync

Introduce WeightUpdater and WeightExporter components (#31148)

2026-07-14 15:53:33 +08:00

constants.py

[AMD][MXFP4] Online MXFP4 quantization 1/N - dense and MOE models w. original BF16 weight (#18005)

2026-06-03 12:55:24 -07:00

environ.py

Integrate pplx a2a backend (#30756)

2026-07-30 15:33:19 -07:00

runtime_context.py

Revert RuntimeContext config-namespace reads/roles (#31813–#31817) (#32100)

2026-07-22 11:52:41 -07:00

server_args_config_parser.py

fix(server): serialize nested dict config values as JSON (#28094)

2026-06-12 19:27:31 -07:00

server_args.py

Integrate pplx a2a backend (#30756)

2026-07-30 15:33:19 -07:00