fzyzcjy df5192cffa Enable fast silu-and-mul-and-quant fused kernel (#11806)

2025-10-30 18:15:39 +08:00

..

batch_invariant_ops

Add Batch‑Invariant RMSNorm (#12144)

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

checkpoint_engine

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

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

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

…

disaggregation

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

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

distributed

[NPU] fix pp_size>1 (#12195)

2025-10-30 11:18:36 +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

[bug] fix router installation to include additional dependency (#12348)

2025-10-29 12:45:18 -07:00

layers

Enable fast silu-and-mul-and-quant fused kernel (#11806)

2025-10-30 18:15:39 +08:00

lora

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

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

managers

[Perf] Optimize multimodal mm_inputs process in scheduler (#11910)

2025-10-30 10:48:10 +08:00

mem_cache

[NPU] fix pp_size>1 (#12195)

2025-10-30 11:18:36 +08:00

metrics

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

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

model_executor

[NPU] fix pp_size>1 (#12195)

2025-10-30 11:18:36 +08:00

model_loader

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

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

models

[Feature] Initial eagle3 support for Deepseek-like models (#12319)

2025-10-30 17:26:09 +08:00

multimodal

model: support NVILA and NVILA Lite (#10399)

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

parser

[fix] added image token as prefix for deepseek-ocr (#12358)

2025-10-30 16:37:03 +08:00

sampling

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

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

speculative

Fix Duplicate Classmethod in spec_info.py (#12354)

2025-10-29 13:57:10 -07:00

tokenizer

…

tracing

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

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

utils

[NPU] fix pp_size>1 (#12195)

2025-10-30 11:18:36 +08:00

weight_sync

…

_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

…

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

…

operations.py

…

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

…