Liangsheng Yin 2812a3c93a [Spec] Unify spec/non-spec decode result handling and overlap relay-payload gating (#29225)

2026-06-25 00:44:38 -07:00

..

arg_groups

Migrate all ServerArgs fields to Annotated style, reduce add_cli_args by ~2400 lines (#28919)

2026-06-22 08:34:37 -07:00

batch_invariant_ops

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

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

batch_overlap

[AMD][bugfix] Place TBO cuda-graph num_token_non_padded buffer on model devices (#28337)

2026-06-20 18:06:22 -07:00

checkpoint_engine

…

compilation

fix(fa3): no NaN embeddings with fa_skip_kv_cache under piecewise CUDA graph (#27343)

2026-06-15 13:46:22 -07:00

configs

[minimax-m3] Split 1/4: sparse attention ops + JIT kernels + config foundation (#28712)

2026-06-22 13:10:43 -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

fix: prevent stale bitmask leakage in LLGuidance grammar backend (#23653)

2026-06-13 13:27:08 -07:00

debug_utils

[Refactor] Introduce sock_send/sock_recv wrappers for zmq IPC (#29012)

2026-06-23 15:54:36 -07:00

disaggregation

Remove Req.extend_logprob_start_len field and make it pure (#27625)

2026-06-25 09:06:14 +08:00

distributed

[AMD] Support triton backend decode context parallel for Qwen3.5 (#25090)

2026-06-23 17:35:56 -07:00

dllm

Inline extend_range accessors and remove the extend_input_len/fill_len properties (#27611)

2026-06-25 08:56:44 +08:00

elastic_ep

[Refactor] Introduce sock_send/sock_recv wrappers for zmq IPC (#29012)

2026-06-23 15:54:36 -07:00

entrypoints

[Cleanup] IPC struct renames, better typing, and SenderWrapper removal (#29214)

2026-06-24 17:25:24 -07:00

eplb

[Perf] Overlap result D2H copy with the next forward step (#29075)

2026-06-24 01:34:02 -07:00

function_call

kimik2_detector fix the normal text detection before tool call. (#25071)

2026-06-24 15:53:22 +08:00

grpc

[gRPC] Extract gRPC servicer into standalone package (#20478)

2026-03-13 09:13:29 -07:00

hardware_backend

[NPU] perf: precompute mamba conv-state track indices once per batch (#29105)

2026-06-25 14:04:56 +08:00

kv_canary

[Spec] Remove deprecated EAGLE v1 DRAFT_EXTEND forward mode (#28129)

2026-06-13 13:00:21 -07:00

layers

[AMD] [GLM5] skip redundant -inf pre-fill of HIP indexer MQA-logits (#28757)

2026-06-24 23:21:39 -07:00

lora

[LoRA] BF16 support + EP cuda-graph crash fix for experimental_sgl_trtllm MoE-LoRA (#28953)

2026-06-24 21:16:50 -07:00

managers

[Spec] Unify spec/non-spec decode result handling and overlap relay-payload gating (#29225)

2026-06-25 00:44:38 -07:00

mem_cache

[Spec] Dissolve EagleDraftInputV2Mixin so spec-info dataclasses hold data only (#29220)

2026-06-24 18:08:30 -07:00

model_executor

fix(runner): prevent eager token buffer under-allocation (#28894)

2026-06-25 14:37:19 +08:00

model_loader

[NVIDIA] Support NVFP4 MoE for DeepSeek-V4 (#25820)

2026-06-21 19:35:14 -07:00

models

[NPU] Fix the DeepSeek-V2-Coder model accuracy issue (#29042)

2026-06-25 09:14:32 +08:00

multimodal

[VLM] Qwen3-VL / Moss-VL ViT preprocessing optimizations (#28940)

2026-06-24 14:36:29 +08:00

multiplex

Move request-ingress methods to SchedulerRequestReceiver (#25610)

2026-05-18 18:31:17 +08:00

observability

[observability] add Ray metric backend wrappers (#26252)

2026-06-17 19:41:53 -07:00

parser

Fix Responses API request handling (#25881)

2026-06-12 14:47:55 -07:00

platforms

Introduce CpuDeviceMixin and CpuSRTPlatform (#26385)

2026-06-17 17:41:17 -07:00

plugins

Multi platform Plugin (#21388)

2026-04-19 17:23:51 -07:00

ray

feat: add SGLANG_RAY_BUNDLE_INDICES for fine-grained Ray bundle index control (#24667)

2026-05-30 02:19:50 -07:00

sampling

Convert SamplingParams to msgspec Struct (#29198)

2026-06-24 13:42:45 -07:00

session

Inline extend_range accessors and remove the extend_input_len/fill_len properties (#27611)

2026-06-25 08:56:44 +08:00

speculative

[Spec] Merge dflash triton kernels into a single dflash.py (#29228)

2026-06-24 18:57:15 -07:00

state_capturer

[fix] Skip routed expert capture for draft model under spec v2 (#26980)

2026-06-24 16:56:07 -07:00

tokenizer

…

utils

Avoid scattered assignment of extend_input_len and fill_len by merging them into Req.extend_range (#27610)

2026-06-25 08:33:28 +08:00

weight_sync

…

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

[BCG][GLM5] perf: BCG support and prefill enhancements (#27053)

2026-06-24 13:13:35 -07:00

runtime_context.py

Add get_parallel(): a structured accessor for parallel-topology state (#28567)

2026-06-17 20:23:43 -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

[Bugfix] Fix inverted PP-disable condition for diffusion LLM inference (#27700)

2026-06-24 20:18:00 -07:00