Liangsheng YinandRatish P 539f772f54 [PD-Disagg] Fully support external DP dispatch w/ PD-disaggregation mode. (#19268) ...

Co-authored-by: Ratish P <114130421+ratish1@users.noreply.github.com>

2026-02-24 19:58:01 -08:00

..

batch_invariant_ops

[Auto Sync] Update batch_invariant_ops.py (20260221) (#19098)

2026-02-20 18:27:31 -08:00

batch_overlap

Revert "[AMD] support two batch overlapping for mori ep #17953" (#19161)

2026-02-23 01:19:23 +08:00

checkpoint_engine

update pre-commit config (#18860)

2026-02-16 00:18:31 +08:00

compilation

Fix PCG MoE Error (#17739)

2026-02-19 16:48:06 +08:00

configs

Whisper model support & /v1/audio/transcriptions endpoint & benchmark (#16983)

2026-02-23 17:28:37 -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

Fix grammar sync across TP ranks (#17100)

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

debug_utils

Support replication axis in dump comparator (#19282)

2026-02-25 09:48:43 +08:00

disaggregation

[PD-Disagg] Fully support external DP dispatch w/ PD-disaggregation mode. (#19268)

2026-02-24 19:58:01 -08:00

distributed

[ROCm] Use unreg path for custom all-reduce during CUDA graph capture (#19162)

2026-02-22 23:27:31 -08:00

dllm

Refactor: observability code cleanup (#17862)

2026-02-24 18:07:29 -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

[PD-Disagg] Fully support external DP dispatch w/ PD-disaggregation mode. (#19268)

2026-02-24 19:58:01 -08:00

eplb

Refactor: observability code cleanup (#17862)

2026-02-24 18:07:29 -08:00

function_call

fix KimiK2Detector regex patterns with re.DOTALL (#19120)

2026-02-21 22:13:08 +08:00

grpc

Refactor: observability code cleanup (#17862)

2026-02-24 18:07:29 -08:00

hardware_backend/npu

support xverse_moe on npu

2026-02-24 14:11:20 +08:00

layers

[NPU] forward_npu uses native impl by default in MultiPlatformOp (#18545)

2026-02-25 09:16:09 +08:00

lora

[Fix] Add lora tied lm head support (for Qwen2.5, Gemma, etc model need) (#18634)

2026-02-19 00:34:51 +08:00

managers

[PD-Disagg] Fully support external DP dispatch w/ PD-disaggregation mode. (#19268)

2026-02-24 19:58:01 -08:00

mem_cache

Refactor: observability code cleanup (#17862)

2026-02-24 18:07:29 -08:00

model_executor

Refactor CUDA graph input buffers with shared buffer pool (#19180)

2026-02-24 02:24:40 -08:00

model_loader

[Quantization] Support config.json quantization_config format, fix exclude_modules matching, and fix KV cache scale loading for Nemotron (#18546)

2026-02-21 16:14:29 +08:00

models

[ROCm] Optimize Deepseek R1 on MI300X (#18242)

2026-02-24 17:01:14 -08:00

multimodal

Whisper model support & /v1/audio/transcriptions endpoint & benchmark (#16983)

2026-02-23 17:28:37 -08:00

multiplex

[Bugfix] fix the display error (inconsistent context) (#17699)

2026-01-31 23:35:11 -08:00

observability

Refactor: observability code cleanup (#17862)

2026-02-24 18:07:29 -08:00

parser

fix: fix the bug blocking completion template application (#17010)

2026-02-24 18:50:08 +08:00

sampling

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

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

speculative

Refactor CUDA graph input buffers with shared buffer pool (#19180)

2026-02-24 02:24:40 -08:00

tokenizer

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

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

utils

Refactor: observability code cleanup (#17862)

2026-02-24 18:07:29 -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

bugfix: The default value of the environment virable is None, so that… (#18309)

2026-02-24 18:10:21 +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

Encoder Global Cache Manager (#16137)

2026-02-24 18:05:43 -08:00