Commits · 751592a6f490aa5d61ee2ab0379e0a8190626403 · OpenDAS / vllm_cscc

"vscode:/vscode.git/clone" did not exist on "995e9a209e68a95ffa03c73f3401472837a4072b"

18 Aug, 2025 3 commits
- Merge remote-tracking branch 'origin/v0.9.2-dev_kvfp8' into v0.9.2-dev · 751592a6
  zhuwenwen authored Aug 18, 2025
  
  751592a6
- Merge remote-tracking branch 'origin/v0.9.2-dev-lzg' into v0.9.2-dev · 787c3715
  zhuwenwen authored Aug 18, 2025
  
  787c3715
- Merge branch 'v0.9.2-dev-wm' into 'v0.9.2-dev' · e7ffe66a
  zhuwenwen authored Aug 18, 2025
```
[fix]修复mtp eager模式下显存占用增加问题

See merge request dcutoolkit/deeplearing/vllm!180
```
  e7ffe66a
16 Aug, 2025 2 commits
- debug and fix erro · 4b4fc875
  lizhigong authored Aug 16, 2025
  
  4b4fc875
- fix mtp accept rate issu · 4829dd84
  lizhigong authored Aug 16, 2025
  
  4829dd84
15 Aug, 2025 4 commits
- [fix]修复mtp eager模式下显存占用增加问题 · 49559d79
  王敏 authored Aug 15, 2025
  
  49559d79
- 修改flashmla的接口 · 299df36d
  yangql authored Aug 15, 2025
  
  299df36d
- add v1 engine + deepseek r1 mtp + zero-overhead scheduler · 5dcaac2f
  lizhigong authored Aug 15, 2025
  
  5dcaac2f
- 增加marlin对cache13的支持，以及新增flash mla的kvcache fp8的支持 · 8f25283a
  yangql authored Aug 15, 2025
  
  8f25283a
14 Aug, 2025 1 commit
- remove seq_lens_tensor of v1 fa · 693d5ed4
  zhuwenwen authored Aug 14, 2025
  
  693d5ed4
13 Aug, 2025 4 commits
- Merge branch 'v0.9.2-dev' of http://10.16.6.30/dcutoolkit/deeplearing/vllm into v0.9.2-dev · e326b648
  zhuwenwen authored Aug 13, 2025
  
  e326b648
- add VLLM_USE_PA_PRINT_PARAM to print v1 fa size · 8a1e7a3d
  zhuwenwen authored Aug 13, 2025
  
  8a1e7a3d
- Merge branch 'v0.9.2-dev-wm' into 'v0.9.2-dev' · 89626cfc
  zhuwenwen authored Aug 13, 2025
```
[feat]优化mtp/eagle的计算逻辑，减少第1层并行解码的计算重复

See merge request dcutoolkit/deeplearing/vllm!178
```
  89626cfc
- [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
12 Aug, 2025 1 commit
- Merge branch 'v0.9.2-dev-lzg' into 'v0.9.2-dev' · a6bf968b
  zhuwenwen authored Aug 12, 2025
```
add v1 engine zero overhead

See merge request dcutoolkit/deeplearing/vllm!177
```
  a6bf968b
11 Aug, 2025 7 commits
- 领导反悔，需要更新beta版本 · 74a444b5
  gaoqiong authored Aug 11, 2025
  
  74a444b5
- 更改回版本为alpha · 02ff7bb1
  gaoqiong authored Aug 11, 2025
  
  02ff7bb1
- add v1 engine zero overhead · 59e80222
  lizhigong authored Aug 11, 2025
  
  59e80222
- update act_and_mul opt · e04730ac
  zhuwenwen authored Aug 11, 2025
  
  e04730ac
- Merge branch 'v0.9.2-dev-pynccl' into 'v0.9.2-dev' · 51fb673c
  zhuwenwen authored Aug 11, 2025
```
已修改        vllm/utils/__init__.py

See merge request dcutoolkit/deeplearing/vllm!176
```
  51fb673c
- add qwen3-coder-480b moe configs · 60e5d686
  zhuwenwen authored Aug 11, 2025
  
  60e5d686
- 已修改 vllm/utils/__init__.py · f2e624d6
  xiabo authored Aug 11, 2025
  
  f2e624d6
10 Aug, 2025 3 commits
- Merge branch 'v0.9.2-dev-wm' into 'v0.9.2-dev' · 8e95b5e2
  gaoqiong authored Aug 10, 2025
```
[fix]修复v1 mtp接受率低的问题

See merge request dcutoolkit/deeplearing/vllm!174
```
  8e95b5e2
- 更改默认的full _cuda_graph启动方式为false · 3dad13fb
  gaoqiong authored Aug 10, 2025
  
  3dad13fb
- [fix]修复v1 mtp接受率低的问题 · 04b61f0e
  王敏 authored Aug 10, 2025
  
  04b61f0e
09 Aug, 2025 3 commits
- update lmslim · dea49b15
  zhuwenwen authored Aug 09, 2025
  
  dea49b15
- Revert "已修改 vllm/utils/__init__.py" · 9ff3592b
  zhuwenwen authored Aug 09, 2025
```
This reverts commit 20b6cf64.
```
  9ff3592b
- Merge branch 'v0.9.2-dev-w8a8-new' into 'v0.9.2-dev' · dcec1430
  zhuwenwen authored Aug 09, 2025
```
V0.9.2 dev w8a8 new

See merge request dcutoolkit/deeplearing/vllm!173
```
  dcec1430
08 Aug, 2025 2 commits
- update fa full_cuda_graph support · 513f17a4
  zhuwenwen authored Aug 08, 2025
  
  513f17a4
- feat:新增VLLM_USE_GLOBAL_CACHE13 设置moe使用全局变量的cache13 · 333104ab
  jujl1 authored Jul 31, 2025
  
  333104ab
07 Aug, 2025 8 commits
- 修改增加SlimQuantW4A8Int8MoEMethod 获取intermediate_size_per_partition 支持 · e92bb9ea
  gaoqiong authored Aug 07, 2025
  
  e92bb9ea
- 修改增加lmslimquant_w4a8量化支持 · 8b1e4ef0
  gaoqiong authored Aug 07, 2025
  
  8b1e4ef0
- Merge branch 'v0.9.2-dev-wm' into 'v0.9.2-dev' · cc6f327a
  zhuwenwen authored Aug 07, 2025
```
[feat]支持mtp模型full_cuda_graph

See merge request dcutoolkit/deeplearing/vllm!172
```
  cc6f327a
- [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 · 89eecc55
  王敏 authored Aug 07, 2025
  
  89eecc55
- [feat]支持mtp模型full_cuda_graph · a1239b53
  王敏 authored Aug 07, 2025
  
  a1239b53
- Merge branch 'v0.9.2-dev-pyncclstream' into 'v0.9.2-dev' · a15d668b
  zhuwenwen authored Aug 07, 2025
```
已修改        vllm/utils/__init__.py

See merge request dcutoolkit/deeplearing/vllm!171
```
  a15d668b
- 已修改 vllm/utils/__init__.py · 20b6cf64
  xiabo authored Aug 07, 2025
  
  20b6cf64
06 Aug, 2025 2 commits
- update VLLM_FLASH_ATTN_V1 to VLLM_USE_FLASH_ATTN_PA · 88dbf92c
  zhuwenwen authored Aug 06, 2025
  
  88dbf92c
- update benchmark_throughput.py · fe657b8b
  zhuwenwen authored Aug 06, 2025
  
  fe657b8b