Commits · cda54326be95fec2430b0f25d8c8e89c900393fc · OpenDAS / vllm_cscc · GitLab

12 Jan, 2026 1 commit
- [feat]添加dp attention功能 · cda54326
  王敏 authored Jan 12, 2026
  
  cda54326
31 Dec, 2025 2 commits
- Revert "增加mtp>1的pad" · d8c36dbf
  yangql authored Dec 31, 2025
```
This reverts commit 059b1961.
```
  d8c36dbf
- 增加mtp>1的pad · 059b1961
  yangql authored Dec 31, 2025
  
  059b1961
29 Dec, 2025 2 commits
- [feat]解决deepep auto模式卡住问题 · cb64c6bc
  王敏 authored Dec 29, 2025
  
  cb64c6bc
- 增加mtp的pad · e60a51b4
  yangql authored Dec 29, 2025
  
  e60a51b4
24 Dec, 2025 2 commits
- [feat]解决EP开启pd_padding大输入卡住问题 · 7439b205
  王敏 authored Dec 24, 2025
  
  7439b205
- [fix]解决EP开启pd_padding大输入卡住问题 · 639a11d2
  王敏 authored Dec 24, 2025
  
  639a11d2
22 Dec, 2025 2 commits
- [fix]1.解决ep sequence parallel优化引入的mtp报错;2.解决共享专家无法和combine overlap问题 · dc01fce4
  王敏 authored Dec 22, 2025
  
  dc01fce4
- [feat]1.优化ep sequence parallel,区分主模型和mtp逻辑;2.ep sequence parallel添加cudagraph... · 62f05dde
  王敏 authored Dec 22, 2025
```
[feat]1.优化ep sequence parallel,区分主模型和mtp逻辑;2.ep sequence parallel添加cudagraph padding到tp_size;3.修复共享专家和deepep combine overlap
```
  62f05dde
18 Dec, 2025 2 commits
- [feat]合入基于deepep的大EP · 13130b89
  王敏 authored Dec 18, 2025
  
  13130b89
- [feat]优化dp attention，减少1次allgather耗时，高吞吐提升明显 · 3e386c3b
  王敏 authored Dec 18, 2025
  
  3e386c3b
27 Nov, 2025 1 commit
- [fix]解决宽松mtp1报错 · deae0a22
  王敏 authored Nov 27, 2025
  
  deae0a22
26 Nov, 2025 1 commit
- [feat]支持宽松mtp · 02b4ec43
  王敏 authored Nov 26, 2025
  
  02b4ec43
25 Aug, 2025 1 commit
- Remove the synchronization introduced by the verification logic when mtp is enabled. · 2fe2d237
  lizhigong authored Aug 25, 2025
  
  2fe2d237
13 Aug, 2025 1 commit
- [feat]优化mtp/eagle的计算逻辑，减少第1层并行解码的计算重复 · 619eb032
  王敏 authored Aug 13, 2025
  
  619eb032
10 Aug, 2025 1 commit
- [fix]修复v1 mtp接受率低的问题 · 04b61f0e
  王敏 authored Aug 10, 2025
  
  04b61f0e
07 Aug, 2025 2 commits
- [feat]支持mtp模型full_cuda_graph · bd58c289
  王敏 authored Aug 07, 2025
  
  bd58c289
- [feat]支持mtp模型full_cuda_graph · a1239b53
  王敏 authored Aug 07, 2025
  
  a1239b53
06 Aug, 2025 2 commits
- [feat]支持mtp模型full_cuda_graph · 9dd945c1
  王敏 authored Aug 06, 2025
  
  9dd945c1
- Revert "Merge remote-tracking branch 'origin/v0.9.2-dev-wm' into v0.9.2-dev" · 0c1cd0f5
  zhuwenwen authored Aug 06, 2025
```
This reverts merge request !169
```
  0c1cd0f5
05 Aug, 2025 2 commits
- [feat]优化mtp相关函数返回类型 · 7e71c143
  王敏 authored Aug 05, 2025
  
  7e71c143
- [feat]1.支持mtp模型 full_cuda_graph; 2.优化mtp拒绝采样 · 8e0ae19d
  王敏 authored Aug 05, 2025
  
  8e0ae19d
31 Jul, 2025 1 commit
- [feat]支持v1 engine mtp cudagraph · fe393be8
  王敏 authored Jul 31, 2025
  
  fe393be8
12 Jul, 2025 1 commit
- [fix]1.优化mtp代码，解决prefix-caching不兼容问题; 2.修复v1 engine从tokenizer config中读取max_model_len导致长文本输入报错 · b16f183c
  王敏 authored Jul 12, 2025
  
  b16f183c
07 Jul, 2025 1 commit
- [feat]支持v1 engine flashmla和mtp同时开启 · ac4dedb1
  王敏 authored Jul 07, 2025
  
  ac4dedb1
06 Jul, 2025 1 commit
- [V1] Support any head size for FlexAttention backend (#20467) · 9fb52e52
  Cyrus Leung authored Jul 07, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  9fb52e52
30 Jun, 2025 1 commit
- [fix]解决v1 engine mtp只能设为1的问题 · 504c262e
  王敏 authored Jun 30, 2025
  
  504c262e
25 Jun, 2025 1 commit
- [V1][Speculative Decoding] Fix DeepSeek MTP (#20022) · 8359f4c8
  cjackal authored Jun 26, 2025
```
Signed-off-by: cjackal <44624812+cjackal@users.noreply.github.com>
```
  8359f4c8
13 Jun, 2025 3 commits
- [CUDA] Enable full cudagraph for FlashMLA (#18581) · 3597b06a
  Luka Govedič authored Jun 13, 2025
```
Signed-off-by: luka <luka@neuralmagic.com>
```
  3597b06a
- [torch.compile] reorganize the cache directory to support compiling multiple models (#19064) · d70bc7c0
  youkaichao authored Jun 13, 2025
```
Signed-off-by: youkaichao <youkaichao@gmail.com>
```
  d70bc7c0
- [Bugfix] Fix EAGLE vocab embedding for multimodal target model (#19570) · c68698b3
  qizixi authored Jun 12, 2025
```
Signed-off-by: qizixi <qizixi@meta.com>
```
  c68698b3
06 Jun, 2025 1 commit
- [Bugfix] Fix EAGLE vocab embedding construction for Llama 70B (#19033) · 3465b87e
  Benjamin Chislett authored Jun 05, 2025
```
Signed-off-by: Benjamin Chislett <benjamin.chislett@centml.ai>
```
  3465b87e
03 Jun, 2025 1 commit
- [Misc] Add SPDX-FileCopyrightText (#19100) · 02f0c7b2
  Simon Mo authored Jun 03, 2025
```
Signed-off-by: simon-mo <simon.mo@hey.com>
```
  02f0c7b2
28 May, 2025 1 commit
- [V1] [Bugfix] eagle bugfix and enable correct lm_head for multimodal (2) (#18781) · 5e13c07d
  RonaldBXu authored May 27, 2025
```
Signed-off-by: Ronald Xu <ronaldxu@amazon.com>
```
  5e13c07d
24 May, 2025 2 commits
- [V1][Spec Decode] Support multi-layer eagle draft model (#18030) · c1e4a405
  qizixi authored May 24, 2025
```
Signed-off-by: qizixi <qizixi@meta.com>
```
  c1e4a405
- [V1][Spec Decode] Small refactors to improve eagle bookkeeping performance (#18424) · d55e446d
  qizixi authored May 23, 2025
```
Signed-off-by: qizixi <qizixi@meta.com>
```
  d55e446d
23 May, 2025 4 commits
- [V1] Support Deepseek MTP (#18435) · 2628a69e
  Jiayi Yao authored May 23, 2025
```
Signed-off-by: Rui Qiao <ruisearch42@gmail.com>
Signed-off-by: YaoJiayi <120040070@link.cuhk.edu.cn>
Co-authored-by: Rui Qiao <ruisearch42@gmail.com>
```
  2628a69e
- Revert "[V1] [Bugfix] eagle bugfix and enable correct lm_head for multimodal (#18034)" (#18600) · fbb13a2c
  Cyrus Leung authored May 23, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  fbb13a2c
- [V1] [Bugfix] eagle bugfix and enable correct lm_head for multimodal (#18034) · 4c611348
  RonaldBXu authored May 23, 2025
```
Signed-off-by: Ronald Xu <ronaldxu@amazon.com>
```
  4c611348
- [V1][Spec Decoding] Use model_loader.get_model() to load models (#18273) · c6b636f9
  Mark McLoughlin authored May 23, 2025
```
Signed-off-by: Mark McLoughlin <markmc@redhat.com>
```
  c6b636f9