sglang

batch_invariant_ops

Add Batch‑Invariant RMSNorm (#12144 )

2025-10-28 21:05:57 -07:00

checkpoint_engine

[Feature] Support loading weights from ckpt engine worker (#11755 )

2025-10-23 09:23:30 -07:00

compilation

[Fix] fix allreduce bug in Piecewise Graph (#12106 )

2025-10-26 21:15:48 +08:00

configs

doc: improve modelopt error description (#12269 )

2025-10-28 20:57:55 -07:00

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

[Feature] New structural tag support (#10691 )

2025-10-20 18:25:58 +08:00

debug_utils

Tiny improve dumper (#11132 )

2025-10-02 17:55:01 +08:00

disaggregation

Fix potential eos bug on decode instance when PD is enabled (#12206 )

2025-10-28 01:29:02 -07:00

distributed

Support releasing CUDA graph memory when paused (#7873 )

2025-10-28 14:40:50 +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

Add continuous_usage_stats support for streaming responses (#12241 )

2025-10-29 10:01:23 +08:00

eplb

EPLB: prefer to use physical experts in the same gpu or node (#10874 )

2025-10-28 21:01:11 -07:00

function_call

[WIP] support MiniMax M2 model (#12129 )

2025-10-26 13:58:54 -07:00

grpc

[grpc] Support gRPC standard health check (#11955 )

2025-10-22 16:59:09 -07:00

layers

Triton fused_moe_kernel support ep moe tuning (#12343 )

2025-10-29 23:16:09 +08:00

lora

Feature/Add GET endpoint to query loaded LoRA adapters (#12229 )

2025-10-28 01:22:00 -07:00

managers

feat: return partial generation results when aborting requests in waiting queue (#11673 )

2025-10-29 22:03:00 +08:00

mem_cache

Revert "[Feature] PD-Multiplexing Context and Scheduler." (#12267 )

2025-10-28 02:00:37 -07:00

metrics

[metrics][EPLB]: Support selected count of physical experts on each GPU (#9825 )

2025-10-28 20:56:19 -07:00

model_executor

EPLB: prefer to use physical experts in the same gpu or node (#10874 )

2025-10-28 21:01:11 -07:00

model_loader

Use Flashinfer TRT-LLM as Llama 4 compatible MoE backend (#11928 )

2025-10-28 10:39:43 -07:00

models

Eagle3 DP attention for Qwen3 MoE (#12002 )

2025-10-29 20:25:17 +08:00

multimodal

model: support NVILA and NVILA Lite (#10399 )

2025-10-26 09:58:09 -07:00

parser

[WIP] support MiniMax M2 model (#12129 )

2025-10-26 13:58:54 -07:00

sampling

Support Thinking Budget (via custom_logit_processor for OpenAI API) [Fix #6572 ] (#11416 )

2025-10-21 16:27:56 +08:00

speculative

Eagle3 DP attention for Qwen3 MoE (#12002 )

2025-10-29 20:25:17 +08:00

tokenizer

move more files under srt/utils (#11285 )

2025-10-09 16:46:15 -07:00

tracing

[Feature] Sglang Tracing: Fine-Grained Tracking for Request Latency - Part 2 (#10804 )

2025-10-28 01:25:46 -07:00

utils

Support releasing CUDA graph memory when paused (#7873 )

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

weight_sync

Tiny move files to utils folder (#11166 )

2025-10-03 22:40:06 +08:00

_custom_ops.py

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

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

constants.py

Support releasing CUDA graph memory when paused (#7873 )

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

custom_op.py

enable llama3.1-8B on xpu (#9434 )

2025-09-07 22:34:20 -07:00

environ.py

feat: return partial generation results when aborting requests in waiting queue (#11673 )

2025-10-29 22:03:00 +08:00

operations_strategy.py

[5/N] MoE Refactor: Update MoE parallelism arguments (#8658 )

2025-08-01 01:20:03 -07:00

operations.py

Add fp4 quantize before all-gather for Flashinfer cutlass MoE DP (max throughput) (#7667 )

2025-08-15 22:08:11 -07:00

server_args_config_parser.py

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

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

server_args.py

fix: llama 4 + trtllm gen + fp8 kv cache incompatibility (#12347 )

2025-10-29 11:31:02 -07:00

single_batch_overlap.py

[hotfix] Incorrect CombineOverlapArgs in SBO (#12230 )

2025-10-28 01:23:06 -07:00

two_batch_overlap.py

[9/N] MoE Refactor: cleanup dispatcher interfaces (#11847 )

2025-10-20 10:11:46 -07:00

warmup.py

[Auto Sync] Update scheduler_profiler_mixin.py, rpd_utils.p... (20250916) (#10494 )

2025-09-16 17:02:20 -07:00