Files

T

Canlin Guoandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> b383e16396 [Bugfix] Reset cached Mamba align metadata on profiling teardown (#54044 )

Signed-off-by: Canlin Guo <canlinguosdu@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

2026-08-30 00:43:20 -07:00

metrics

[Model Runner V2] Support num NaNs in logits (#30187 )

2025-12-09 10:00:49 -08:00

[Perf] Avoid h2d copies from non-pinned CPU tensors (#54299 )

2026-08-29 12:06:30 +08:00

model_states

[K3 Perf] Optimize k3 mamba metadata preparation, 6.6~7.6x kernel performance improvement (#52388 )

2026-08-25 11:00:28 -04:00

pool

[Pooling] Improve BGE-M3 sync pooling throughput by up to 3.13% (#53464 )

2026-08-24 08:33:32 -04:00

sample

[Bugfix][Model Runner V2][Spec Decode] Decouple the draft's gumbel noise stream from the target's (#54282 )

2026-08-29 09:58:29 -07:00

spec_decode

[Bugfix][Model Runner V2][Spec Decode] Decouple the draft's gumbel noise stream from the target's (#54282 )

2026-08-29 09:58:29 -07:00

__init__.py

GPU Model Runner V2 (#25266 )

2025-11-21 08:20:55 -08:00

async_utils.py

[Feature] Mask Replay (#49577 )

2026-08-13 08:44:31 +00:00

attn_utils.py

[Bugfix][MRV2] Isolate sleep-mode KV allocations (#53508 )

2026-08-27 15:08:28 -07:00

block_table.py

[Bugfix][MRV2] Isolate sleep-mode KV allocations (#53508 )

2026-08-27 15:08:28 -07:00

buffer_utils.py

[Core] Ensure memory is pinned prior to async h2d copy (#45424 )

2026-06-20 20:02:24 -07:00

cp_utils.py

[Spec decode] Support Kimi-K3 DCP with DSpark (#52188 )

2026-08-17 13:08:45 -07:00

cudagraph_utils.py

[Bugfix] Reset cached Mamba align metadata on profiling teardown (#54044 )

2026-08-30 00:43:20 -07:00

dp_utils.py

[Model Runner V2][Spec Decode] Skip DP sync before EAGLE/MTP draft prefill (#53694 )

2026-08-26 20:59:35 -07:00

ec_connector.py

[EC Connector] Added Build Connector Worker Meta for EC Connector (#49585 )

2026-08-15 15:40:38 -07:00

eplb_utils.py

[Elastic EP] Reduce eager-mode reconfiguration downtime (#51885 )

2026-08-19 02:13:06 +00:00

input_batch.py

[Spec Decode] DSpark confidence-scheduled verification (#47808 )

2026-08-12 12:27:48 -07:00

kv_connector.py

[Core][KV Connector] Start async KV loads after the forward launch when no sync loads are scheduled (#53333 )

2026-08-28 21:58:00 +00:00

lora_utils.py

[BugFix][MRV2] Don't create dummy requests longer than max_model_len (#49751 )

2026-07-27 02:04:20 +00:00

model_runner.py

[Nixl][PD] DCP support for MLA models (#50611 )

2026-08-29 16:03:53 +00:00

pcp_manager.py

BugFix(PCP): use persistent input buffers for PIECEWISE CUDA graphs (#53515 )

2026-08-26 05:58:13 +00:00

pp_utils.py

[XPU] Fix PP accuracy on XPU device (#47253 )

2026-07-07 09:17:27 +08:00

README.md

GPU Model Runner V2 (#25266 )

2025-11-21 08:20:55 -08:00

shutdown.py

[MRV2] Add shutdown() method (#41297 )

2026-05-02 21:06:30 -07:00

states.py

[Model Runner V2][Spec Decode] Add multi-layer MTP speculator (#48892 )

2026-07-30 15:54:15 -07:00

structured_outputs.py

[Bugfix][Spec Decode][Structured Output] DSpark: fix the grammar bitmask mapping when the draft budget is zero (#52436 )

2026-08-16 11:09:23 +00:00

warmup.py

[Refactor][Model Runner V2][Multimodal] Move the encoder-only path out of the shared runner (#53176 )

2026-08-22 05:11:44 +00:00

README.md

[Experimental] Model Runner V2

This directory contains the new model runner which is under active development. Ping Woosuk Kwon for any changes.