sglang

batch_invariant_ops

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

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

batch_overlap

[Feature] Xiaomi MiMo-V2-Flash day0 support (#15207 )

2025-12-19 11:40:07 +08:00

checkpoint_engine

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

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

compilation

Custom All Reduce for Piecewise Cuda Graph (#15356 )

2025-12-25 23:55:15 +08:00

configs

[NemotronH] Add latent MoE support (#16227 )

2026-01-02 22:08:58 +08:00

connector

[Lint] Add python/sglang to ruff F401 checks and remove unused imports in files (#11685 )

2025-10-17 16:49:46 -07:00

constrained

Adding tool calling and reasoning parser support for Intern-S1 (#14866 )

2025-12-16 00:01:58 -08:00

debug_utils

Support simple schedule simulator (#16344 )

2026-01-03 22:05:40 +08:00

disaggregation

Move log_prefill_stats_late to correct location in PP mode (#15946 )

2025-12-31 12:00:24 +08:00

distributed

[distributed] Clean up MoE groups in destroy_model_parallel (#15345 )

2026-01-02 14:45:14 +08:00

dllm

Tiny extract ModelRunnerOutput (#15400 )

2025-12-18 22:18:45 +08:00

elastic_ep

[2/N] Added the core structure of elastic EP and the eplb algorithm with faulty rank (#10606 )

2025-10-22 01:13:31 -07:00

entrypoints

[Auto Sync] Update engine.py (20260102) (#16317 )

2026-01-02 17:29:40 -08:00

eplb

Support EPLB balancedness prometheus metric without GPU->CPU synchronize (#15401 )

2025-12-18 22:24:23 +08:00

function_call

Fix DeepSeekV31's structural tag trigger (#13394 )

2025-12-31 21:13:52 +08:00

grpc

[model-gateway] Add classification model support infrastructure (#16061 )

2025-12-29 08:34:05 -08:00

hardware_backend/npu

[DeepSeek v3.2] opt Context Parallelism: support fused moe, multi batch and fp8 kvcache (#13959 )

2026-01-02 23:49:14 +08:00

layers

[Performance] Optimze the performance of Qwen25VL (#15640 )

2026-01-02 23:15:36 -08:00

lora

[LoRA] Torch native backend: rework implementation and updated tests (#15187 )

2025-12-30 11:48:51 +08:00

managers

Fix incorrect running batch size in prefill stats (#15941 )

2026-01-03 20:49:54 +08:00

mem_cache

[HiCache] Add PP Support with suffix pp rank (#15175 )

2026-01-03 13:49:21 +08:00

metrics

Tiny add explanations to realtime token metric labels (#16009 )

2026-01-03 12:36:55 +08:00

model_executor

[Performance] Optimze the performance of Qwen25VL (#15640 )

2026-01-02 23:15:36 -08:00

model_loader

Download missing shards in model weights files when not in CI (#16211 )

2025-12-31 20:42:34 +08:00

models

[NPU] Adapt qwen3-next W8A8 on NPU (#16164 )

2026-01-03 19:41:20 +08:00

multimodal

[VLM] feat: true on policy for vlm + fsdp (#14636 )

2026-01-01 16:54:39 -08:00

multiplex

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

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

parser

fix: prevent points regex from matching checkpoints/endpoints (#15120 )

2025-12-17 12:38:18 -08:00

sampling

Support returning customized info (#16144 )

2026-01-02 15:32:18 +08:00

speculative

Eagle: GPT-OSS Eagle v2 support (#14920 )

2025-12-30 14:23:07 -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

Tiny support whitelisted headers in request logging (#16342 )

2026-01-03 20:10:58 +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

Add host tensor allocator for memory_pool_host and support Mooncake standalone storage (#14873 )

2025-12-28 16:28:50 -08:00

server_args_config_parser.py

Fix parse args from file(#13911 ) (#14085 )

2026-01-01 11:37:33 +08:00

server_args.py

Tiny add --log-requests-target (#16338 )

2026-01-03 17:28:27 +08:00