Yuhui Liang ca029e816b Fix missing idle-batch handling in prepare_mlp_sync_batch_raw (#25404)

2026-05-24 23:53:54 -07:00

..

arg_groups

[Refactor] Rename NSA → DSA: user-facing aliases, file/class/import rename (#25821)

2026-05-20 00:18:04 -07:00

batch_invariant_ops

Register aten::rms_norm and aten::mm.dtype in batch invariant mode (#24459)

2026-05-05 17:21:34 -07:00

batch_overlap

feat(model_runner): remove pool/backend refs from ForwardBatch via ForwardContext (#25983)

2026-05-21 14:01:49 -07:00

checkpoint_engine

update pre-commit config (#18860)

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

compilation

Support piecewise CUDA graph with NSA (#23351)

2026-05-22 14:39:50 -07:00

configs

Clean up server startup log noise (#26205)

2026-05-24 14:35:15 -07:00

connector

[connector] Add Azure Blob Storage connector (az:// and *.blob.core.windows.net URLs) (#23995)

2026-05-11 14:20:22 -07:00

constrained

feat(constrained): two-phase reasoning grammar + --enable-strict-thinking (#23953)

2026-05-07 14:21:51 -07:00

debug_utils

Add e2e test with log snapshot in dumper grafter (#24513)

2026-05-06 17:00:13 +08:00

disaggregation

[PD] Consolidate shared logic into common backend (#25979)

2026-05-23 10:41:20 +08:00

distributed

[PD] Un-blacklist mooncake sessions when probe succeeds (#25287)

2026-05-20 11:16:28 +08:00

dllm

perf: migrate Req token-id storage to array.array('q') in Scheduler (#25098)

2026-05-22 10:51:07 -07:00

elastic_ep

[6/N] (Elastic EP) Recover failed ranks (#15771)

2026-04-28 00:44:26 -07:00

entrypoints

feat(kv-events): expose structured KV-event publisher block on /server_info (#25844)

2026-05-23 01:59:45 +08:00

eplb

[observability] add ServerArgs.stat_loggers for pluggable metrics backend (#24610)

2026-05-24 22:41:28 +08:00

function_call

Add MiniCPM5 tool call parser for XML-style function calls (#25600)

2026-05-22 23:09:40 +08:00

grpc

[gRPC] Extract gRPC servicer into standalone package (#20478)

2026-03-13 09:13:29 -07:00

hardware_backend

[XPU] fix correctness issue of GDN triton kernel for XPU (#26065)

2026-05-25 13:18:11 +08:00

layers

[perf][spec decoding] Skip common_template in TRTLLMMLAMultiStepDraftBackend init (#26241)

2026-05-24 21:36:16 -07:00

lora

[lora] Fix overlap loading for cancelled requests (#25413)

2026-05-25 15:18:36 +09:00

managers

Fix missing idle-batch handling in prepare_mlp_sync_batch_raw (#25404)

2026-05-24 23:53:54 -07:00

mem_cache

[Bug Fix][HiCache] TreeNode.get_prefix_hash_values @lru_cache can return mutated list (#26177)

2026-05-25 11:15:10 +08:00

model_executor

Clean up server startup log noise (#26205)

2026-05-24 14:35:15 -07:00

model_loader

loader: yield filtered MTP weights lazily to avoid OOM hang on multi-layer EAGLE (#25748)

2026-05-20 12:33:54 +08:00

models

[VLM] feat: accept grid_thws from preprocessed metadata for kimi (#26149)

2026-05-25 09:10:25 +08:00

multimodal

[VLM] fix: fix only the grids from last split mm item is collected for qwen-vl (#26094)

2026-05-25 09:09:46 +08:00

multiplex

Move request-ingress methods to SchedulerRequestReceiver (#25610)

2026-05-18 18:31:17 +08:00

observability

[observability] add ServerArgs.stat_loggers for pluggable metrics backend (#24610)

2026-05-24 22:41:28 +08:00

parser

[RL][TITO] Preserve whitespace in reasoning parser outputs (#24251)

2026-05-20 19:45:09 +00:00

platforms

[Refactor] Rename NSA → DSA: user-facing aliases, file/class/import rename (#25821)

2026-05-20 00:18:04 -07:00

plugins

Multi platform Plugin (#21388)

2026-04-19 17:23:51 -07:00

ray

Bundle Scheduler rank/size fields into a frozen ParallelState (#25444)

2026-05-16 09:23:51 +08:00

sampling

Reject repetition_penalty=0 in SamplingParams.verify() (#24874)

2026-05-12 21:25:23 -07:00

session

[UnifiedRadixCache] Fix HiCache load back start node (#25088)

2026-05-14 13:18:12 +08:00

speculative

[Spec] fix EAGLE v2 verify metadata init order on non-cuda-graph path (#26244)

2026-05-24 18:49:33 -07:00

state_capturer

[Session R3] Add routed_experts_start_len for absolute routing slice control (#24851)

2026-05-10 10:04:43 -07:00

tokenizer

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

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

utils

Add --disable-attn-tp-gather opt-out for model-managed SP (#26047)

2026-05-24 15:53:51 -07:00

weight_sync

[RL] Allow passing tensors of different dtypes for FlattenedTensorBucket (#13413)

2025-11-19 17:24:48 +08:00

constants.py

Clean up detokenizer and remove dead multimodal_gen code (#21588)

2026-03-28 21:44:40 -07:00

environ.py

[AMD] Dsv4/pr1 fix run time issue (#25898)

2026-05-23 16:04:14 -07:00

server_args_config_parser.py

[ConfigArgumentMerger] Improve ConfigArgumentMerger compatibility with external callers (#17051)

2026-01-16 15:32:42 +08:00

server_args.py

Add --disable-attn-tp-gather opt-out for model-managed SP (#26047)

2026-05-24 15:53:51 -07:00