zouyida2052 90aca84e60 fix bug when max_seqs=14 in mtp=2 scenario and raise error when cudagraph_capture_sizes can't be an integer multiple of uniform_decode_query_len (#3909) ...

### What this PR does / why we need it?
1. Revert [bugfix for mtp in
fullgraph](0948483642)
and support it when vllm supports
2. raise error when cudagraph_capture_sizes can't be an integer multiple
of uniform_decode_query_len
3. bugfix when max_num_seqs=14 in mtp=2 scenario

---------

Signed-off-by: zouyida2052 <zouyida2002@gmail.com>

2025-10-31 09:25:06 +08:00

..

attention

[BugFix] Fix mlapo accuracy problem related with weight processing. (#3857)

2025-10-30 00:35:50 +08:00

compilation

[v0.11.0][Fix] Prevent memory leak in MLA decode graph (#3743) (#3774)

2025-10-27 16:00:20 +08:00

core

[BugFix][Cherry-pick] Cherry-pick PR 3675 to v0.11.0-dev (#3732)

2025-10-25 09:41:51 +08:00

device_allocator

[Misc]Clean up useless import from vllm (#2049)

2025-07-28 16:01:59 +08:00

distributed

[bugfix_v0.11.0]cancel tokenize for layerwise_proxy (#3913)

2025-10-30 23:55:04 +08:00

eplb

[CI]Add EPLB CI. (#3568)

2025-10-21 22:58:02 +08:00

lora

[Bugfix][LoRA] Fix forward error and shape mismatch when using LoRA (#3153)

2025-09-28 17:30:50 +08:00

models

Revert "[Feat] Shared expert dp for deepseek and deepseek_mtp (#3495)" (#3586)

2025-10-21 22:24:30 +08:00

multistream

[Quickfix] update CachedRequestState as NewRequestData changed (#2367)

2025-08-15 07:35:27 +08:00

ops

[BugFix]Fix group list type of mc2. (#3890)

2025-10-30 21:44:14 +08:00

patch

[cherry-pick]【main】patch sched_yield (#3648) (#3687)

2025-10-24 00:24:58 +08:00

quantization

[v0.11.0][bugfix] Add 'layer_type' param to get_pergroup_param() for compatibility (#3684)

2025-10-23 21:26:50 +08:00

sample

Drop 0.10.2 (#3284)

2025-10-09 10:28:38 +08:00

spec_decode

unify logic between aclgraph and torchair (#3602)

2025-10-22 21:55:06 +08:00

torchair

fix bug when max_seqs=14 in mtp=2 scenario and raise error when cudagraph_capture_sizes can't be an integer multiple of uniform_decode_query_len (#3909)

2025-10-31 09:25:06 +08:00

worker

fix bug when max_seqs=14 in mtp=2 scenario and raise error when cudagraph_capture_sizes can't be an integer multiple of uniform_decode_query_len (#3909)

2025-10-31 09:25:06 +08:00

__init__.py

[Refactor] Adapt deepseek-v3.2 to vllm 0.11.0 (#3432)

2025-10-15 17:48:58 +08:00

ascend_config.py

[main] support cpu binding (#3546)

2025-10-21 09:17:03 +08:00

ascend_forward_context.py

[v0.11.0][Perf] Delete redundant operations in model_runner and forward_context (#3775)

2025-10-29 15:58:53 +08:00

cpu_binding.py

[main] support cpu binding (#3546)

2025-10-21 09:17:03 +08:00

envs.py

[Feat] Flash comm allgher ep (#3334)

2025-10-15 19:36:32 +08:00

meta_registration.py

Fix the bugs about operator registration by PyTorch Dispatcher (#2786)

2025-09-13 11:58:52 +08:00

platform.py

fix bug when max_seqs=14 in mtp=2 scenario and raise error when cudagraph_capture_sizes can't be an integer multiple of uniform_decode_query_len (#3909)

2025-10-31 09:25:06 +08:00

utils.py

fix bug when max_seqs=14 in mtp=2 scenario and raise error when cudagraph_capture_sizes can't be an integer multiple of uniform_decode_query_len (#3909)

2025-10-31 09:25:06 +08:00