Mohammad Miadh Angkad ad7125a431 [Bugfix] Fix DeepSeek V4 MTP HC state handling (#42320) ...

Signed-off-by: Mohammad Miadh Angkad <176301910+mmangkad@users.noreply.github.com>
(cherry picked from commit f1cc7aad3c)

2026-05-14 21:28:34 -07:00

..

assets

[Examples][last/6] Resettle examples. (#41084)

2026-05-07 19:42:12 -07:00

benchmarks

Fix spec decode benchmark metrics (#41916)

2026-05-07 09:23:21 -07:00

compilation

[Bugfix] Fix SP pass for multimodal models and PP+SP residual handling (#33322)

2026-05-09 19:44:16 -07:00

config

[Performance] Make safetensors checkpoint prefetch settings configurable (#41499)

2026-05-10 15:55:15 +00:00

device_allocator

Auto-disable expandable_segments around cumem memory pool (#40812)

2026-04-27 09:37:22 +08:00

distributed

[KV Transfer] Add MooncakeStoreConnector for KV cache offloading via Mooncake distributed store (#40900)

2026-05-13 02:02:44 -07:00

engine

[Performance] Make safetensors checkpoint prefetch settings configurable (#41499)

2026-05-10 15:55:15 +00:00

entrypoints

[Frontend] Return rendered prompt text in chat completion response (#42052)

2026-05-11 13:53:39 +08:00

inputs

feat: Enable prompt_embeds Content Part Support in vLLM Chat Completions API (#40720)

2026-05-01 10:05:55 +08:00

ir

[vLLM IR] 2/N fused_add_rms_norm and maybe_inplace overload (#36823)

2026-05-01 23:41:15 -04:00

kernels

[Kernel][Helion] Optimize Helion config parsing latency (#40850)

2026-05-07 20:27:34 -07:00

logging_utils

[vLLM IR] 1/N Implement IR skeleton and rms_norm op (#33825)

2026-03-31 22:15:05 -04:00

lora

[Bugfix] Fuse Qwen3.5 in_qkvz_proj forwarding with LoRA enabled (#37912)

2026-05-10 10:59:02 +00:00

model_executor

[Bugfix] Fix DeepSeek V4 MTP HC state handling (#42320)

2026-05-14 21:28:34 -07:00

multimodal

[Model Runner V2] support qwen35 / mamba hybrid model (#35520)

2026-05-07 09:31:05 -07:00

parser

[Refactor] Consolidate required/named tool_choice streaming into DelegatingParser (#41876)

2026-05-07 09:50:59 -07:00

platforms

[MLA Attention Backend] Add TOKENSPEED_MLA backend for DSR1/Kimi K25 prefill + decode on Blackwell (#41778)

2026-05-14 00:59:51 -07:00

plugins

[Frontend][4/n] Improve pooling entrypoints | pooling. (#39153)

2026-04-09 10:09:45 +00:00

profiler

[Refactor] Clean up log once scope="local" (#40540)

2026-04-22 16:42:43 -04:00

ray

…

reasoning

Fix: Nemotron 3 rescue whitespace-only final_content, not just None (#41846)

2026-05-10 02:07:58 +00:00

renderers

[Bugifx] Missing Renderer for fastokens mode (#41984)

2026-05-07 23:45:14 -07:00

third_party

…

tokenizers

tokenizer: Add fastokens support (#41741)

2026-05-07 22:49:42 +08:00

tool_parsers

[Bugfix] Preserve leading/trailing whitespace in GLM non-streaming tool parser (#42026)

2026-05-08 21:49:15 -07:00

tracing

…

transformers_utils

[Spec Decode] Add Gemma4 MTP speculative decoding support (#41745)

2026-05-06 22:39:29 +08:00

triton_utils

[Feat][CPU] Enable Gated DeltaNet Attention (Qwen 3.5 / 3.6) (#41025)

2026-05-07 12:57:09 +08:00

usage

replace cuda_device_count_stateless() to current_platform.device_count() (#37841)

2026-03-31 22:32:54 +08:00

utils

[CPU] Fix spec decode kernel signatures for synthetic mode compatibility (#41932)

2026-05-10 12:07:15 +00:00

v1

[MLA Attention Backend] Add TOKENSPEED_MLA backend for DSR1/Kimi K25 prefill + decode on Blackwell (#41778)

2026-05-14 00:59:51 -07:00

vllm_flash_attn

[Attention] use diff kv backend for mimo v2 flash (#40045)

2026-04-24 11:25:55 +00:00

__init__.py

…

_aiter_ops.py

[ROCm][Perf] Add Fused Shared Expert (FSE) support for Qwen3-Next (#39280)

2026-05-08 15:38:00 -04:00

_custom_ops.py

Use CU_MEMCPY_SRC_ACCESS_ORDER_ANY for batch KV cache swaps (#39306)

2026-05-10 05:57:09 +03:00

_xpu_ops.py

[XPU] remove q/k/v force contiguous for flash_attn (#40356)

2026-05-09 01:19:05 +00:00

beam_search.py

[Misc] Reorganize inputs (#35182)

2026-03-25 10:22:54 -07:00

collect_env.py

[XPU]Enhance environment collection for Intel XPU and optimize layout (#35698)

2026-04-13 12:51:46 +08:00

connections.py

[CI] Add retry with 4x backoff to HTTP fetches for transient failures (#37218)

2026-03-19 19:00:20 +00:00

env_override.py

fix: default TILELANG_CLEANUP_TEMP_FILES=1 to avoid shared /tmp conflicts (#41486)

2026-05-07 19:59:00 -07:00

envs.py

[XPU] use xpu topk topp sample kernel (#39285)

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

exceptions.py

[openapi server] log exception in exception handler(2/N) (#36201)

2026-03-10 20:16:30 -07:00

forward_context.py

[MoE Refactor] Remove MoE DP chunking (#39107)

2026-04-14 09:48:05 -04:00

logger.py

[Log] Log once in local node by default (#37568)

2026-03-19 12:04:59 -07:00

logits_process.py

…

logprobs.py

…

model_inspection.py

…

outputs.py

[Core] Replace routing replay with device cache and async D2H pipeline (#39917)

2026-05-07 11:24:56 -07:00

pooling_params.py

[Frontend] Remove frontend pooling multi task support. (#37861)

2026-04-21 12:27:44 +00:00

py.typed

…

sampling_params.py

[Model Runner V2] Add logprob_token_ids support (#40559)

2026-05-01 10:02:03 -07:00

scalar_type.py

…

scripts.py

…

sequence.py

[Model Runner V2][Minor] Simplify PP logic (#38031)

2026-03-24 13:57:17 -07:00

tasks.py

[Frontend] Remove frontend pooling multi task support. (#37861)

2026-04-21 12:27:44 +00:00

version.py

…