XueLiang Yang e390742c59 Fix KV Offloading + MLA AssertionError by using num_kv_heads=1 in cpu… (#37536) ...

Signed-off-by: xueliangyang-oeuler <yxl546827391@gmail.com>
Co-authored-by: xueliangyang-oeuler <yxl546827391@gmail.com>

2026-03-19 12:05:07 +00:00

..

attention

[Bugfix] Zero-init MLA attention output buffers to prevent NaN from CUDA graph padding (#37442)

2026-03-19 00:28:37 +00:00

core

Make KV connector metadata build overridable via plugin (#37336)

2026-03-17 21:29:06 +00:00

engine

[bug] Fix deadlock with pause resume and collective_rpc (#37024)

2026-03-19 01:49:02 +00:00

executor

[bugfix][async scheduling] fix extra cuda context in device 0 with EP/DP (#37449)

2026-03-18 18:32:30 +00:00

kv_offload

Fix KV Offloading + MLA AssertionError by using num_kv_heads=1 in cpu… (#37536)

2026-03-19 12:05:07 +00:00

metrics

[Bugfix] Expand quantization method support in perf metrics (#37231)

2026-03-18 23:54:19 +00:00

pool

[Perf] Optimize token_embed for pooling models, 1.0% token throughput improvement (#37347)

2026-03-19 10:27:51 +08:00

sample

[Perf] Optimize top-k search in apply_top_k_top_p_triton sampler (#37225)

2026-03-17 11:35:17 -07:00

spec_decode

[Bugfix] dtype mismatch in ngram gpu propose (#37246)

2026-03-17 05:19:55 +00:00

structured_output

[MISC] fix pin_memory=torch.cuda.is_available(), use is_pin_memory_available (#37415)

2026-03-19 09:23:24 +00:00

worker

Support temporal compression for Nemotron-3-VL videos (#36808)

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

__init__.py

[V1] AsyncLLM Implementation (#9826)

2024-11-11 23:05:38 +00:00

cudagraph_dispatcher.py

[UX][Startup] Account for CUDA graphs during memory profiling (#30515)

2026-03-07 13:49:23 -08:00

kv_cache_interface.py

[BUGFIX][Mamba][Qwen3.5] Zero freed SSM cache blocks on GPU (#35219)

2026-03-10 03:32:20 -07:00

outputs.py

[KVConnector] Support worker -> scheduler metadata (#31964)

2026-03-11 17:36:37 +00:00

request.py

[Deprecation][1/2] Remove items deprecated in v0.18 (#36470)

2026-03-09 03:43:23 -07:00

serial_utils.py

[Bug] Fix Failure in /v1/chat/completions/render for Multimodal Requests (https://github.com/vllm-project/vllm/issues/35665) (#35684)

2026-03-14 14:10:11 +00:00

utils.py

[Frontend][Core] Re-add shutdown timeout - allowing in-flight requests to finish (#36666)

2026-03-13 12:10:06 -07:00