Commits · 9d5187eb68ef31f5fb5d63f4e3b4862fb186aad2 · OpenDAS / vllm_cscc

27 Jun, 2025 2 commits
- [Model] Add support for MiniMaxM1ForCausalLM (shares architecture with MiniMaxText01ForCausalLM) · 95d58c0d
  zhuwenwen authored Jun 27, 2025
  
  95d58c0d
- update cutlass fa and pa · 4ddb5447
  zhuwenwen authored Jun 27, 2025
  
  4ddb5447
24 Jun, 2025 1 commit
- update SUPPORT_TC · fdda4d82
  zhuwenwen authored Jun 24, 2025
  
  fdda4d82
23 Jun, 2025 3 commits
- update common.py · 2da6fd72
  zhuwenwen authored Jun 23, 2025
  
  2da6fd72
- update bench run error · f246ee95
  zhuwenwen authored Jun 23, 2025
  
  f246ee95
- fix moe run error · 10184690
  zhuwenwen authored Jun 23, 2025
  
  10184690
21 Jun, 2025 4 commits
- support ndhwc of qwen2-vl and qwen2.5-vl · 5f71c3f8
  zhuwenwen authored Jun 21, 2025
  
  5f71c3f8
- 添加moe smquant量化支持模块 · 7996b363
  zhuwenwen authored Jun 21, 2025
  
  7996b363
- only K100_AI can use cutlass prefix cache · 14945681
  zhuwenwen authored Jun 21, 2025
  
  14945681
- skip is_rocm_aiter_moe_enabled and remove qwen unused int8 code · bbd14169
  zhuwenwen authored Jun 21, 2025
  
  bbd14169
18 Jun, 2025 4 commits
- 分离kmai 120cu与120cu w8a8-int8优化config共享 · 1b3e4da4
  gaoqiong authored Jun 18, 2025
  
  1b3e4da4
- use custom_all_reduce when bs*hidden_size<=256*4096 (bw) · d59f30d4
  zhuwenwen authored Jun 18, 2025
  
  d59f30d4
- release opt1 · 1a280da1
  zhuwenwen authored Jun 18, 2025
  
  1a280da1
- 暂时去掉fused_moe中量化kernel的bug修复 · 35dbdc41
  王敏 authored Jun 18, 2025
  
  35dbdc41
17 Jun, 2025 3 commits
- [fix]1.修复medusa、scorer等并行解码单测；2.修复moe... · 8db76782
  王敏 authored Jun 17, 2025
```
[fix]1.修复medusa、scorer等并行解码单测；2.修复moe kernel单测问题，优化代码；3.修复rejection_sampler中test_compare_nonflashinfer_backend单测问题
```
  8db76782
- fix zero overhead bug when kvcache oom and schedule · bd1e64d6
  lizhigong authored Jun 17, 2025
  
  bd1e64d6
- 修复int4选取config的bug · 44e87dde
  yangql authored Jun 17, 2025
  
  44e87dde
16 Jun, 2025 1 commit
- 分离fuse moe awq算子到lmslim上 · b3312eec
  yangql authored Jun 16, 2025
  
  b3312eec
13 Jun, 2025 6 commits
- remove pkg_resources · a9b29641
  zhuwenwen authored Jun 13, 2025
  
  a9b29641
- update triton_mla.py · e046b382
  zhuwenwen authored Jun 13, 2025
  
  e046b382
- 增加fused_experts_impl_int8的接入 · 483acdc4
  xuxz authored Jun 13, 2025
  
  483acdc4
- [fix]修复并行解码eagle和mlp相关单测问题 · 08cb5d8f
  王敏 authored Jun 13, 2025
  
  08cb5d8f
- 解耦w8a8 的gemm以及moe算子相关实现 · 942368c7
  gaoqiong authored Jun 13, 2025
  
  942368c7
- [New Model]: Support Qwen3 Embedding & Reranker · 9bcbaafc
  zhuwenwen authored Jun 13, 2025
  
  9bcbaafc
12 Jun, 2025 7 commits
- add bw qwen3 moe configs · 2f6f5bb3
  zhuwenwen authored Jun 12, 2025
  
  2f6f5bb3
- 修复qwen3-moe的awq配置导致fp16加载错误，修复dpsk-moe手写算子首字耗时增加问题 · 99ffef47
  zhuwenwen authored Jun 12, 2025
  
  99ffef47
- Update int8_utils.py · 550a1e5e
  zhuwenwen authored Jun 12, 2025
  
  550a1e5e
- set self.max_seq_len_to_capture = self.max_model_len · d8b9028d
  zhuwenwen authored Jun 12, 2025
  
  d8b9028d
- update qwen2&2.5-vl prefill interface and fix pixtral run error · 4fd5389b
  zhuwenwen authored Jun 12, 2025
  
  4fd5389b
- add kvcache fp8 · 504a12b8
  zhuwenwen authored Jun 12, 2025
  
  504a12b8
- add VLLM_TBO_DECODE_BS to support and setting the min bs on tbo decode · aa906d98
  lizhigong authored Jun 12, 2025
  
  aa906d98
10 Jun, 2025 9 commits
- [Misc] Slight improvement of the BNB (#19418) · b6553be1
  Jee Jee Li authored Jun 10, 2025
```
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Co-authored-by: Isotr0py <2037008807@qq.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
```
  b6553be1
- [BugFix][FlashInfer] Fix attention backend interface mismatch with unexpected... · 467bef18
  Rachel Guo authored Jun 10, 2025
```
[BugFix][FlashInfer] Fix attention backend interface mismatch with unexpected keyword `use_irope` (#19134)
Signed-off-by: Yunqiu Guo <guorachel@meta.com>
```
  467bef18
- Revert "[v1] Add fp32 support to v1 engine through flex attn" (#19404) · 5f1ac1e1
  Isotr0py authored Jun 10, 2025
  
  5f1ac1e1
- Automatically bind CPU OMP Threads of a rank to CPU ids of a NUMA node. (#17930) · 9368cc90
  Louie Tsai authored Jun 09, 2025
```
Signed-off-by: Tsai, Louie <louie.tsai@intel.com>
Co-authored-by: Li, Jiang <bigpyj64@gmail.com>
```
  9368cc90
- [Core] Batch multi modal input using pinned memory (#19169) · 319cb1e3
  Lukas Geiger authored Jun 10, 2025
```
Signed-off-by: Lukas Geiger <lukas.geiger94@gmail.com>
```
  319cb1e3
- [Bugfix] Fix modelscope token passed in (#19389) · 1efef716
  Li Wang authored Jun 10, 2025
```
Signed-off-by: wangli <wangli858794774@gmail.com>
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
```
  1efef716
- [Core] Use tuple for kv cache group block ids (#19175) · 646d62f6
  Nick Hill authored Jun 09, 2025
```
Signed-off-by: Nick Hill <nhill@redhat.com>
```
  646d62f6
- [Frontend] Add tqdm_leave_pbar to control progress bar visibility (#19357) · 6cd4ae8a
  Reid authored Jun 10, 2025
```
Signed-off-by: reidliu41 <reid201711@gmail.com>
Co-authored-by: reidliu41 <reid201711@gmail.com>
```
  6cd4ae8a
- Use xla flag to improve the quantized model performance (#19303) · 9af6d22e
  XiongfeiWei authored Jun 09, 2025
```
Signed-off-by: Xiongfei Wei <isaacwxf23@gmail.com>
```
  9af6d22e