sglang

arg_groups

[DeepSeek V4] Enable FlashMLA sparse prefill by default (#29775 )

2026-07-01 13:50:05 -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

[Model] Add HrmTextForCausalLM (Hierarchical Reasoning Model - Text) (#27887 )

2026-06-30 13:32:32 +08:00

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(spec): track current_token in ReasonerGrammarObject (#29102 )

2026-06-26 16:43:35 -07:00

debug_utils

Convert IPC dataclasses to msgspec.Struct with opt-in msgpack transport (#28688 )

2026-06-26 12:04:03 -07:00

disaggregation

[disagg] Fix KV-event publisher port collision under pure data parallelism (#29211 )

2026-07-01 10:57:34 -07:00

distributed

Extract reusable VMM shareable-handle helpers from register_graph_inputs (#29621 )

2026-07-01 14:02:00 -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

[passthrough] engine: zstd request-body decompression + header overrides (#29684 )

2026-06-30 23:48:56 -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

feat(mem_cache): unified memory pool for hybrid Mamba / SWA models (#29678 )

2026-07-01 13:21:59 -07:00

kv_canary

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

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

layers

feat(short-conv): shared ShortConvAttnBackend for ZAYA1 CCA + LFM2 short conv (#29867 )

2026-07-01 20:08:00 -07:00

lora

Fuse the DSA (V3.2, GLM-5.x) indexer Q/K paths into single kernels (#27705 )

2026-06-27 14:29:56 -07:00

managers

feat(mem_cache): unified memory pool for hybrid Mamba / SWA models (#29678 )

2026-07-01 13:21:59 -07:00

mem_cache

Fix MambaPool.clear_slots OOM by replacing expand-based tensor allocation with scalar zeroing (#27923 )

2026-07-02 09:26:55 +08:00

model_executor

[Spec] Enable FlashInfer autotune for spec draft (#29595 )

2026-07-01 13:36:07 -07:00

model_loader

Fix bounded checkpoint prefetching and buffered drop-cache handling (#29156 )

2026-06-29 21:49:16 +00:00

models

feat(short-conv): shared ShortConvAttnBackend for ZAYA1 CCA + LFM2 short conv (#29867 )

2026-07-01 20:08:00 -07:00

multimodal

model: support nvidia/LocateAnything-3B (#28958 )

2026-06-30 00:16:42 +08:00

multiplex

Move request-ingress methods to SchedulerRequestReceiver (#25610 )

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

observability

Fix KV event publisher bind conflict under PP (#29044 )

2026-06-26 12:50:49 +08:00

parser

model: support baidu unlimited-ocr (#29186 )

2026-06-27 23:36:19 +08:00

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 IPC dataclasses to msgspec.Struct with opt-in msgpack transport (#28688 )

2026-06-26 12:04:03 -07:00

session

feat: first-class session identity in SGLang (#29436 )

2026-06-28 07:16:38 -07:00

speculative

[Spec] Enable FlashInfer autotune for spec draft (#29595 )

2026-07-01 13:36:07 -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

[Fix]: Inline H2D during CUDA graph capture to avoid stream isolation in Offloader (#29166 )

2026-06-30 17:49:58 -07:00

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

[DeepSeek V4] Enable FlashMLA sparse prefill by default (#29775 )

2026-07-01 13:50:05 -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

feat(mem_cache): unified memory pool for hybrid Mamba / SWA models (#29678 )

2026-07-01 13:21:59 -07:00