Alison Shao 52c8a3632a Fix missing StandardCombineInput import in BF16 flashinfer_trtllm MoE (#19400) ...

Co-authored-by: Alison Shao <alisonshao@MacBook-Pro-D2W773R9CD.local>

2026-02-26 18:32:08 -08:00

..

batch_invariant_ops

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

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

batch_overlap

[Bugfix] Add rids to the batch filtering for two batch overlap (#19418)

2026-02-26 06:57:25 -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

[NPU][feature adapt]remote load weight feature adp npu (#17968)

2026-02-27 10:10:36 +08:00

constrained

Fix grammar sync across TP ranks (#17100)

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

debug_utils

Enhance displaying and debuggability in dump comparator (#19466)

2026-02-27 09:06:19 +08:00

disaggregation

[Minor] Rename misleading chunked to reusing in ReqToTokenPool.alloc() (#19465)

2026-02-26 16:24:03 -08:00

distributed

refactor: extract device-to-backend mapping into get_default_distributed_backend (#19202)

2026-02-24 23:50:26 -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

[misc] use ORJSONResponse in http-server generate (#19191)

2026-02-25 21:26:25 -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

[NPU] bugfix for model Qwen3-Coder-Next at weight shape transpose for npu. (#18700)

2026-02-25 15:46:20 +08:00

layers

Fix missing StandardCombineInput import in BF16 flashinfer_trtllm MoE (#19400)

2026-02-26 18:32:08 -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

[Spec V2] Support specV2 for mamba hybrid attention (#18808)

2026-02-27 00:36:01 +08:00

mem_cache

[Minor] Rename misleading chunked to reusing in ReqToTokenPool.alloc() (#19465)

2026-02-26 16:24:03 -08:00

model_executor

[Spec V2] Support specV2 for mamba hybrid attention (#18808)

2026-02-27 00:36:01 +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

[AMD] Fix weight load shape mismatch for amd dsr1 0528 mxfp4 (#19425)

2026-02-26 17:11:59 -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

[Spec V2] Support specV2 for mamba hybrid attention (#18808)

2026-02-27 00:36:01 +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

refactor linear attention backend (#18622)

2026-02-25 23:02:44 +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

[NPU][feature adapt]remote load weight feature adp npu (#17968)

2026-02-27 10:10:36 +08:00