ovidiusm 498d8d0680 NixlKVManager optimizations (#17654) ...

Signed-off-by: Ovidiu Mara <ovidium@nvidia.com>

2026-02-06 00:25:23 +08:00

..

batch_invariant_ops

fix: add bias when enable mm fallback variant (#17690)

2026-01-28 09:50:49 +08:00

batch_overlap

[MiMoV2Flash] [feat]: support two batch overlap (#17634)

2026-02-02 14:03:57 -08:00

checkpoint_engine

…

compilation

Fix torch.__version__ for PEP440 (#15682)

2026-01-29 11:55:13 -08:00

configs

model: support interns1-pro (#18145)

2026-02-05 00:22:44 +08:00

connector

…

constrained

Fix grammar sync across TP ranks (#17100)

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

debug_utils

Support multiple engines for router simulation in schedule simulator (#16368)

2026-01-04 11:46:05 +08:00

disaggregation

NixlKVManager optimizations (#17654)

2026-02-06 00:25:23 +08:00

distributed

Make sure we always disable symm memory without dp padding (#18129)

2026-02-04 19:58:28 +08:00

dllm

[DLLM] Implement initial dynamic batching for diffusion LLM (#14883)

2026-01-17 16:48:15 +08:00

elastic_ep

…

entrypoints

entrypoint: support passing spaces_between_special_tokens per request (#17939)

2026-02-04 22:18:36 +08:00

eplb

Re-introduce the unit test of test_mooncake_ep_small (#16019)

2026-01-12 17:01:24 +08:00

function_call

Gigachat 3 tool parser and tests (#14765)

2026-02-02 22:28:34 -08:00

grpc

[gRPC] Add GetLoads RPC for comprehensive load metrics (#17087)

2026-01-14 12:05:10 -08:00

hardware_backend/npu

Moving _alloc_extend_naive out of npu allocator (#18200)

2026-02-04 02:09:55 -08:00

layers

[Kernel] Add JIT apply_rope_with_cos_sin_cache_inplace (#18155)

2026-02-05 21:49:37 +08:00

lora

throw error if got adapter with added_tokens (#18046)

2026-02-05 23:55:43 +08:00

managers

fix kimi k2.5's moe gemm config init (#18064)

2026-02-04 16:59:01 -08:00

mem_cache

optimize get_topk_ragged by fusing get k and k_scale triton kernel (#16043)

2026-02-04 19:59:41 +08:00

metrics

[HiCache] feat: Add detailed cache hit breakdown for HiCache in sglext and Prometheus metrics (#17648)

2026-02-03 11:45:35 -08:00

model_executor

model: support Step-3.5-Flash (#18084)

2026-02-03 00:40:07 +08:00

model_loader

[Perf] Use safetensors load_file in multithread loader (#18124)

2026-02-02 23:21:13 -08:00

models

[piecewise graph]: support MiniMax-M2 (#18217)

2026-02-04 23:24:38 -08:00

multimodal

model: support interns1-pro (#18145)

2026-02-05 00:22:44 +08:00

multiplex

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

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

parser

model: support Step-3.5-Flash (#18084)

2026-02-03 00:40:07 +08:00

sampling

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

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

speculative

model: support Step-3.5-Flash (#18084)

2026-02-03 00:40:07 +08:00

tokenizer

…

tracing

SGLang Tracing: Improve root span attributes (#17008)

2026-01-30 16:02:05 +08:00

utils

[XPU] Integrate MoE and minor improvements in XPU attention backend (#13561)

2026-02-04 23:09:59 -08:00

weight_sync

…

constants.py

…

environ.py

[DeepGemm] Add a flag for fast warmup (#18111)

2026-02-04 14:12:13 +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

model: support Step-3.5-Flash (#18084)

2026-02-03 00:40:07 +08:00