b8zhong 24a25ffa20 [Piecewise CUDA Graph] Support ModelOpt FP4 (#13101)

2025-11-16 11:03:19 +08:00

..

batch_invariant_ops

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

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

checkpoint_engine

…

compilation

Piecewise Cuda Graph Support for gpt-oss model (#13045)

2025-11-15 09:28:00 +08:00

configs

[Quantization] Support Quark Dense + MoE FP8 & FP8 PTPC (#10485)

2025-11-13 08:16:00 -08:00

connector

…

constrained

…

debug_utils

Add script to create a model with fewer layers for debugging (#13284)

2025-11-14 22:13:27 +08:00

disaggregation

[Feature] Spec-Overlap supporting DP-ATTN; PD-Disaggregation; npugraph mode (#12443)

2025-11-15 21:51:07 +08:00

distributed

[Auto Sync] Update pynccl_wrapper.py, environ.py, registry.... (20251111) (#13097)

2025-11-13 15:58:47 -08:00

elastic_ep

…

entrypoints

[Fix] Register custom ops only if they exist (#13321)

2025-11-15 17:48:17 -08:00

eplb

Fix nan in global scaling factor for large scale nvfp4 EP (#13162)

2025-11-12 15:32:23 -08:00

function_call

[Tool Call] Steamline function arguments when tool_choice="auto" for deepseekv31_detector (#11589)

2025-11-13 23:06:02 -08:00

grpc

Add timing metrics for requests (#12646)

2025-11-05 23:07:16 -08:00

layers

[Piecewise CUDA Graph] Support ModelOpt FP4 (#13101)

2025-11-16 11:03:19 +08:00

lora

[Ascend] LoRA: adding Ascend LoRA backend with using kernels from sgl_kernel_npu (#12288)

2025-11-12 16:20:34 -08:00

managers

[Fix] Register custom ops only if they exist (#13321)

2025-11-15 17:48:17 -08:00

mem_cache

[optimize] Provide Usrbio compilation and installation commands (#12329)

2025-11-15 23:06:24 +08:00

metrics

refine stdout logging codes (#13015)

2025-11-12 00:16:14 +08:00

model_executor

[Feature] Spec-Overlap supporting DP-ATTN; PD-Disaggregation; npugraph mode (#12443)

2025-11-15 21:51:07 +08:00

model_loader

Tiny refactor condition to requant scale ue8m0 (#13286)

2025-11-15 16:36:00 +08:00

models

Opt tp: tp attn support tp reduce scattered input (#10568)

2025-11-15 02:08:12 -08:00

multimodal

fix: duplicate resize images logic of qwen-vl series models (#12458)

2025-11-12 18:08:40 +08:00

multiplex

[Feature] PD-Multiplexing Context and Scheduler, lazy import spatial. (#12275)

2025-11-01 00:40:01 +08:00

parser

add Kimi k2 reasoning parser (#12702)

2025-11-06 00:37:54 +08:00

sampling

…

speculative

[Feature] Spec-Overlap supporting DP-ATTN; PD-Disaggregation; npugraph mode (#12443)

2025-11-15 21:51:07 +08:00

tokenizer

…

tracing

[minor] remove debug code in python/sglang/srt/compilation/weak_ref_tensor_jit.py (#13235)

2025-11-14 12:22:24 -08:00

utils

[Fix] Register custom ops only if they exist (#13321)

2025-11-15 17:48:17 -08:00

weight_sync

…

_custom_ops.py

move all get_stream in sgl_kernel to c++ to reduce the launch overhead (#12521)

2025-11-02 13:15:05 -08:00

constants.py

…

custom_op.py

…

environ.py

[RL] Allow bypassing /health check (#13320)

2025-11-15 16:33:54 +08:00

operations_strategy.py

…

operations.py

Register allgather/reducescatter buffers with symm memory (#12572)

2025-11-04 17:11:36 -08:00

server_args_config_parser.py

…

server_args.py

[Ascend][Feat] Add Ascend sampling backend (#12692)

2025-11-15 21:59:58 +08:00

single_batch_overlap.py

…

two_batch_overlap.py

Minor code cleanup / improvement for PREBUILT_EXTEND mode (#12948)

2025-11-10 11:59:54 +08:00

warmup.py

…