zhangheng e4b708d3e9 [Spec V2] Support specV2 for mamba hybrid attention (#18808) ...

Co-authored-by: Yi Zhong <207368749+vincentzed@users.noreply.github.com>
Co-authored-by: yizhang2077 <1109276519@qq.com>
Co-authored-by: Hanming Lu <hanming@x.ai>

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

..

4-gpu-models

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

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

8-gpu-models

Fix nightly Mistral-Large-3 NVFP4 accuracy threshold (#19402)

2026-02-26 15:59:11 +08:00

amd

[AMD] add testcases for Qwen3 235b Instruct 2507 models (#18805)

2026-02-25 02:17:52 -08:00

ascend

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

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

attention

update pre-commit config (#18860)

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

backends

…

bench_fn

[Benchmark] Fix generated_shared_prefix attribute naming and remove args dependency (#19363)

2026-02-25 18:45:54 -08:00

constrained_decoding

…

core

Use ephemeral nccl port via get_free_port() (#18009)

2026-02-16 00:32:47 +08:00

debug_utils

Support multi-step alignment and pipeline integration in dump comparator (#19378)

2026-02-26 10:23:22 +08:00

disaggregation

[CI] feat: add early exit to wait_for_server when process dies (#18602)

2026-02-13 16:46:09 -08:00

distributed

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

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

dllm

Revert "Add SDAR model support" (#19032)

2026-02-19 16:03:56 -08:00

ep

chore: bump mooncake version to 0.3.9 (#18316)

2026-02-07 17:30:01 +08:00

eval

Fix eval tests not capturing server launch failures (#18886)

2026-02-18 07:59:03 +08:00

function_call

…

hicache

[Benchmark] Remove re-exports from bench_serving.py (#19130)

2026-02-21 14:30:30 -08:00

kernels

[DSv32] [GLM5] Improve Model Quality by Avoiding FP32 Precision Loss in weights_proj (#19041)

2026-02-22 16:20:51 +08:00

layers

Feat/add fi selective state update kernel call (#18070)

2026-02-19 16:56:06 +08:00

lora

Move lora request validation to tokenizer_manager from server (#18962)

2026-02-19 21:03:19 +08:00

metrics

Refactor: observability code cleanup (#17862)

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

mla

fix: add --cuda-graph-max-bs to DSV3 FA3 FP8 KV cache test (#19307)

2026-02-25 10:27:31 -08:00

model_loading

…

models

[PCG] add pcg CI test for gpt-oss model (#18961)

2026-02-24 15:32:34 -08:00

moe

Merge stage-c-test-large-4-gpu suites into partitioned suites (#18325)

2026-02-06 15:32:33 -08:00

openai_server

fix tool handling in OpenAIServingChat (#18996)

2026-02-21 22:07:09 +08:00

ops

[AMD] Support --enable-aiter-allreduce-fusion on AMD GPUs (#13747)

2026-02-24 23:11:55 -08:00

parser

…

perf

[AMD] Update aiter to v0.1.10.post2 (#18423)

2026-02-08 22:08:24 -08:00

piecewise_cuda_graph

Tiny rename cuda_graph tests to piecewise_cuda_graph (#19128)

2026-02-21 13:46:01 -08:00

profiling

…

quant

test: add test for Modelopt FP8 on SM90 (#18463)

2026-02-16 00:29:37 +08:00

radix_cache

[RadixTree][4/N Refactor]: Move available_and_evictable_str to individual radix cache classes (#17852)

2026-02-19 17:03:15 +08:00

rl

[CI] Move test_load_lora_from_tensor test to H100 (#18797)

2026-02-13 21:28:00 +08:00

rotary

feat: Support mrope_section with rope_type: "yarn" (#13313)

2026-02-14 22:51:44 +08:00

sampling

Fix flaky penalty tests by using higher temperature for effect comparison (#18380)

2026-02-12 21:08:37 +08:00

scheduler

Extract dumper and prefill delayer tests common utils (#18857)

2026-02-15 18:33:23 +08:00

spec

Add timeout abort kits for normal / eagle. (#18815)

2026-02-13 17:57:30 -08:00

stress

…

tokenizer

…

utils

chore: fix some typos (#18577)

2026-02-10 20:47:41 -08:00

vlm

…

test_hybrid_dp_ep_tp_mtp.py

…

test_srt_backend.py

…