Henry e5edf222cd [WIP]enable mxfp8 on nvidia sm120 (#19112) ...

Co-authored-by: Your Name <you@example.com>

2026-03-01 19:06:43 -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 "[SGL] sync patch: Remove sync points, prefill cudagraph for DP, disable cache reset in mem check (#19190)" (#19581)

2026-02-28 19:46:47 -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

Revert "[SGL] sync patch: Remove sync points, prefill cudagraph for DP, disable cache reset in mem check (#19190)" (#19581)

2026-02-28 19:46:47 -08:00

constrained

…

debug_utils

Support data parallel attention in dump comparator (#19602)

2026-03-01 10:51:21 +08:00

disaggregation

[HiCache] Re-land spec v2 + decode KV cache offloading compatibility (#19615)

2026-03-01 13:58:31 -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

[4/N] (Elastic EP) Back up Expert Weights in DRAM (#17374)

2026-02-27 15:59:13 +08:00

entrypoints

[Bugfix] Fix KeyError: 'prompt_tokens' when streaming requests are aborted (#19514)

2026-02-28 20:21:35 -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] bugs fix for Deepseek models (#19544)

2026-02-28 17:26:15 +08:00

layers

[WIP]enable mxfp8 on nvidia sm120 (#19112)

2026-03-01 19:06:43 -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

[Bugfix] Add missing auto_create_handle_loop to communicator methods (#19610)

2026-03-01 15:00:05 -08:00

mem_cache

Revert "[SGL] sync patch: Remove sync points, prefill cudagraph for DP, disable cache reset in mem check (#19190)" (#19581)

2026-02-28 19:46:47 -08:00

model_executor

Revert "[SGL] sync patch: Remove sync points, prefill cudagraph for DP, disable cache reset in mem check (#19190)" (#19581)

2026-02-28 19:46:47 -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

[fix typo] expert_indicies -> expert_indices (#19627)

2026-03-01 17:37:34 -08:00

multimodal

[vlm][internVL] Support processor and embedding inputs for InternVL (#19127)

2026-02-26 22:46:48 -08:00

multiplex

…

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

…

speculative

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

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

tokenizer

…

utils

Refactor: observability code cleanup (#17862)

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

weight_sync

…

constants.py

…

environ.py

[4/N] (Elastic EP) Back up Expert Weights in DRAM (#17374)

2026-02-27 15:59:13 +08:00

server_args_config_parser.py

…

server_args.py

feat: add FA4 SM90 paged KV decode support & update attention docs (#18442)

2026-03-02 09:12:19 +08:00