sglang

batch_invariant_ops

Support fast gemm when in batch invariant DeepGEMM fallback (#13259 )

2025-11-15 16:34:15 +08:00

batch_overlap

Overlap shared experts with deepep dispatch for single batch overlap on Blackwell (#17289 )

2026-01-21 02:56:55 +08:00

checkpoint_engine

[Ckpt Engine] feat: new sglang entrypoint support for update (#12216 )

2025-10-30 10:39:27 +08:00

compilation

[Piecewise] Support PCG weak_ref_tensor cuda kernel on AMD (#17291 )

2026-01-20 14:05:32 -08:00

configs

Lazy import torchao (#17626 )

2026-01-22 22:04:51 -08:00

connector

…

constrained

Fix grammar sync across TP ranks (#17100 )

2026-01-15 18:38:01 -08:00

debug_utils

Support multiple engines for router simulation in schedule simulator (#16368 )

2026-01-04 11:46:05 +08:00

disaggregation

Skip mm feature pool init to avoid EPD OOM (#16388 )

2026-01-23 10:53:45 +08:00

distributed

[CPU][Fix CI] Solidate torch version for sgl-kernel-cpu and fix device orientation error (#17460 )

2026-01-22 14:04:50 +08:00

dllm

[DLLM] Implement initial dynamic batching for diffusion LLM (#14883 )

2026-01-17 16:48:15 +08:00

elastic_ep

…

entrypoints

Add return routed experts to the completions and chat/completions endpoints (#17434 )

2026-01-23 12:12:36 -08:00

eplb

Re-introduce the unit test of test_mooncake_ep_small (#16019 )

2026-01-12 17:01:24 +08:00

function_call

model: step3-vl-10b (#17513 )

2026-01-22 23:15:08 +08:00

grpc

[gRPC] Add GetLoads RPC for comprehensive load metrics (#17087 )

2026-01-14 12:05:10 -08:00

hardware_backend/npu

[NPU]support model MiniCPM3-4B for npu (#16866 )

2026-01-24 08:25:12 +08:00

layers

Move fa4 from sgl-kernel to jit kernel (#17353 )

2026-01-24 15:25:03 +08:00

lora

add documentation example for LoRA overlap loading and cleanup unused function (#17464 )

2026-01-24 15:33:16 +08:00

managers

add documentation example for LoRA overlap loading and cleanup unused function (#17464 )

2026-01-24 15:33:16 +08:00

mem_cache

Bugfix: Writing to storage when write-back method is chosen (#14718 )

2026-01-22 15:08:25 -08:00

metrics

Code clean up for fp8 quantization (#16982 )

2026-01-13 12:38:39 -08:00

model_executor

add documentation example for LoRA overlap loading and cleanup unused function (#17464 )

2026-01-24 15:33:16 +08:00

model_loader

[Fix] GLM 4.7 + NVFP4 + MTP (#17166 )

2026-01-21 21:34:18 +08:00

models

Use attn tp group in embedding for more models (#17570 )

2026-01-24 13:37:44 +08:00

multimodal

Skip mm feature pool init to avoid EPD OOM (#16388 )

2026-01-23 10:53:45 +08:00

multiplex

Fix deprecation warning of sre_parse for python 3.13 (#16247 )

2025-12-31 21:25:09 -08:00

parser

VLM: enhance VL embedding model with video input support and revise warm-up strategy (#16635 )

2026-01-08 22:12:01 +08:00

sampling

Revert "feat: reduce constrained-decoding overhead in TP" (#16845 )

2026-01-10 11:39:38 +08:00

speculative

Evict swa kv cache during decoding (#17220 )

2026-01-19 22:36:52 +08:00

tokenizer

[Piecewise Cuda Graph] rename, refactor and add more logging (#13675 )

2025-11-21 13:28:38 +08:00

tracing

SGLang Tracing: fix attribute errors (header extraction & bootstrap span closing) (#15693 )

2025-12-27 22:44:31 -06:00

utils

[MUSA][1/N] sglang.check_env (#16959 )

2026-01-23 14:41:17 -08:00

weight_sync

[RL] Allow passing tensors of different dtypes for FlattenedTensorBucket (#13413 )

2025-11-19 17:24:48 +08:00

constants.py

Support releasing CUDA graph memory when paused (#7873 )

2025-10-28 14:40:50 +08:00

environ.py

Support symmetric memory pre-allocation to avoid fragmentation (#17089 )

2026-01-23 17:57:04 +08:00

server_args_config_parser.py

[ConfigArgumentMerger] Improve ConfigArgumentMerger compatibility with external callers (#17051 )

2026-01-16 15:32:42 +08:00

server_args.py

[DLLM] Remove cuda graph batch size limitation (#17458 )

2026-01-23 09:52:39 -08:00