sglang

arg_groups

Allow CUDA VMM feature transport with the Rust frontend (#39347 )

2026-09-14 17:09:49 -07:00

batch_invariant_ops

keeping router GEMM in fp32 for deterministic inference (DeepSeek V3/V4) (#38176 )

2026-09-15 16:16:09 +08:00

batch_overlap

[CP V1 Deprecation 3.5/5] Deprecate HIP/NPU/MUSA prefill CP and remove legacy implementation (#38293 )

2026-09-07 15:39:34 -07:00

beam_search

[CI][RFC] Replace black-jupyter with ruff-format (#37210 )

2026-09-02 19:46:08 -07:00

checkpoint_engine

[XPU] Make checkpoint_engine worker device-agnostic (#32382 )

2026-09-11 09:56:39 +08:00

compilation

[CI][RFC] Replace black-jupyter with ruff-format (#37210 )

2026-09-02 19:46:08 -07:00

configs

[Qwen3.8] Enable NVIDIA NVFP4 on DGX Spark with file-backed PLE and PDL router fix (#39126 )

2026-09-13 16:23:41 +08:00

connector

[CI][RFC] Replace black-jupyter with ruff-format (#37210 )

2026-09-02 19:46:08 -07:00

constrained

Support non-strict GLM47 tool calls with EBNF constraints (#38890 )

2026-09-15 16:41:48 +08:00

debug_utils

[XPU] Adapt device agnostic API usage (#32093 )

2026-09-11 10:03:12 +08:00

disaggregation

[Intra-node PD][DSV4] Pack all layers into one batch for INTRA_NODE_NVLINK path (#38984 )

2026-09-15 21:08:33 +08:00

distributed

Fix mooncake scale joiner groups (#38503 )

2026-09-15 19:07:25 +08:00

dllm

[CI][RFC] Replace black-jupyter with ruff-format (#37210 )

2026-09-02 19:46:08 -07:00

elastic_ep

[Config] Round 6.5: a namespace declares what it derives, next to what it derives it from (#38113 )

2026-09-06 21:44:24 -07:00

entrypoints

Support non-strict GLM47 tool calls with EBNF constraints (#38890 )

2026-09-15 16:41:48 +08:00

eplb

[CI][RFC] Replace black-jupyter with ruff-format (#37210 )

2026-09-02 19:46:08 -07:00

function_call

Support non-strict GLM47 tool calls with EBNF constraints (#38890 )

2026-09-15 16:41:48 +08:00

hardware_backend

[NPU] Fix device mismatch in SWA mask for DSpark verify graph capture (#39353 )

2026-09-14 19:36:33 +08:00

kv_canary

Fix KV-canary workspace accounting after graph capture (#38596 )

2026-09-10 16:10:12 -07:00

layers

[EPLB] Drop defensive getattr for ep_dispatch_algorithm (#31804 )

2026-09-15 16:14:41 +08:00

lora

[LoRA] Support MoE in full and breakable prefill CUDA graphs (#38578 )

2026-09-11 14:40:02 -07:00

managers

[Fix] Aggregate all IPC weight update responses (#39534 )

2026-09-14 23:25:46 -07:00

mem_cache

[Fix][DCP] Localize widened KV ids in MLA retraction CPU backup/restore (#39487 )

2026-09-15 13:33:41 +08:00

model_executor

[profiler] Label draft-runner steps DRAFT and target verify VERIFY in step spans (#38630 )

2026-09-15 00:57:05 -07:00

model_loader

[model-loader] Split weight loading from postprocessing (#34981 )

2026-09-14 17:10:19 -07:00

models

[qwen 3.8 next] reuse old cuda stream instead of endlessly creating streams (#39474 )

2026-09-15 01:36:14 -07:00

multimodal

[Fix][Qwen-VL] Normalize <image> sentinel on artifact fast path (#39278 )

2026-09-15 13:19:59 +08:00

multiplex

[Scheduler] Unify per-iteration request intake into ingest_requests() (#38389 )

2026-09-07 20:02:02 -07:00

observability

[observability] Fix missing e2e/decode/inference latency span attributes (#37789 )

2026-09-08 22:50:55 +08:00

parser

Force reasoning mode for GLM-5.3 chat templates (#39227 )

2026-09-15 13:16:26 +08:00

platforms

[NPU][Bugfix] Disable pinned memory to fix DeepSeek-V2 DP-attention hang (#39423 )

2026-09-15 14:59:31 +08:00

plugins

[CI][RFC] Replace black-jupyter with ruff-format (#37210 )

2026-09-02 19:46:08 -07:00

ray

[Config] One writer for the declaration stash; no exception to the write seal (#38752 )

2026-09-09 19:22:06 -07:00

rust_extensions

[Rust TreeCore] Harden runtime and CI parity (#37303 )

2026-09-09 10:40:22 +08:00

rust_server

[Rust] Use Dynamo native renderers when chat templates are missing (#38939 )

2026-09-14 17:29:10 -07:00

sampling

Support non-strict GLM47 tool calls with EBNF constraints (#38890 )

2026-09-15 16:41:48 +08:00

session

Scope prefetch cache state to the request attempt (#39318 )

2026-09-14 15:03:03 -07:00

speculative

[Unified Cache][7/N] Support MTP, EAGLE, and DSpark draft KV caches in the external linker (#37914 )

2026-09-13 18:22:06 +08:00

state_capturer

state_capturer: pin the exact host-cache size via mmap + cudaHostRegister (#37285 )

2026-09-03 16:20:59 -07:00

tokenizer

[CI][RFC] Replace black-jupyter with ruff-format (#37210 )

2026-09-02 19:46:08 -07:00

utils

[profiler] Label draft-runner steps DRAFT and target verify VERIFY in step spans (#38630 )

2026-09-15 00:57:05 -07:00

weight_cache

[Config] Round 6.4: the runtime reads the bags, not the record (#38049 )

2026-09-06 21:41:46 -07:00

weight_sync

Introduce WeightUpdater and WeightExporter components (#31148 )

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

constants.py

Refactor server_args constants and layout (#36676 )

2026-08-27 12:16:16 -07:00

environ.py

[MM] Add flag to force Kimi image preprocessing onto CPU (#39148 )

2026-09-14 21:19:39 -07:00

runtime_context.py

Add a provider hook for prefill-buffer ceilings (#39182 )

2026-09-12 21:37:10 -07:00

server_args.py

[Perf] Optimize Qwen3-VL unique-image serving on H100 (#36411 )

2026-09-12 15:16:59 +08:00