Commits · 08c4bafa439c82c9093ead81f38c58ee61503961 · OpenDAS / vllm_cscc

26 Mar, 2025 1 commit
- add auto finish and restart thread when finish generate and start new generate · 08c4bafa
  lizhigong authored Mar 26, 2025
  
  08c4bafa
25 Mar, 2025 2 commits
- Merge remote-tracking branch 'origin/v0.7.2-dev' into v0.7.2_zero_overhead · ca4ec0ce
  lizhigong authored Mar 25, 2025
  
  ca4ec0ce
- debug and fix some error about outputs · 0be169ad
  lizhigong authored Mar 25, 2025
  
  0be169ad
24 Mar, 2025 10 commits
- seq.data._effective_length+=1 · 18b9f67c
  guanyu1 authored Mar 24, 2025
  
  18b9f67c
- detok补充 · 6d0996e9
  guanyu1 authored Mar 24, 2025
  
  6d0996e9
- Merge branch 'v0.7.2-dev-wm' into 'v0.7.2-dev' · ae0ed592
  zhuwenwen authored Mar 24, 2025
```
[fix]修复fused_moe.py中fused_moe接口未初始化moe_ep_size导致的deekseek等模型报错

See merge request dcutoolkit/deeplearing/vllm!89
```
  ae0ed592
- [fix]修复fused_moe.py中fused_moe接口未初始化moe_ep_size导致的deekseek等模型报错 · 4b7f382f
  王敏 authored Mar 24, 2025
  
  4b7f382f
- detok修改 · 3a49f3d2
  guanyu1 authored Mar 24, 2025
  
  3a49f3d2
- detok修改 · 052059d9
  guanyu1 authored Mar 24, 2025
  
  052059d9
- update triton_mla.py · a54eca71
  zhuwenwen authored Mar 24, 2025
  
  a54eca71
- use two thread in step too improve first tokens · 2344d22e
  lizhigong authored Mar 24, 2025
  
  2344d22e
- add mla tuning configs of k100-ai · 30e0b082
  zhuwenwen authored Mar 24, 2025
  
  30e0b082
- add mla tuning script and configs · 13b1dcfe
  zhuwenwen authored Mar 24, 2025
  
  13b1dcfe
19 Mar, 2025 1 commit
- debug zero overhead · b78549c2
  lizhigong authored Mar 19, 2025
  
  b78549c2
18 Mar, 2025 2 commits
- fix llm_engine to zero_overhead · 7d867671
  lizhigong authored Mar 18, 2025
  
  7d867671
- update benchmarks · b95d1275
  zhuwenwen authored Mar 18, 2025
  
  b95d1275
17 Mar, 2025 3 commits
- update version · d9e67e78
  zhuwenwen authored Mar 17, 2025
  
  d9e67e78
- sampler修改 · 08c2298a
  guanyu1 authored Mar 17, 2025
  
  08c2298a
- zero overhead engine update · 9bd32639
  lizhigong authored Mar 17, 2025
  
  9bd32639
15 Mar, 2025 12 commits
- update fused_moe.py · 8a7c4451
  zhuwenwen authored Mar 15, 2025
  
  8a7c4451
- update qwen2-vl and qwen2.5-vl layout · 37ef8dd3
  zhuwenwen authored Mar 15, 2025
  
  37ef8dd3
- set llama and qwen tn layout · b845811e
  zhuwenwen authored Mar 15, 2025
  
  b845811e
- add cust2 · 72de3bf9
  zhuwenwen authored Mar 15, 2025
  
  72de3bf9
- Merge remote-tracking branch 'origin/0.7.2-zhangshao' into v0.7.2-pa · c9d28c80
  zhuwenwen authored Mar 15, 2025
  
  c9d28c80
- 调整pa tc和非tc调用关系 · 0d2ccb8d
  zhangshao authored Mar 15, 2025
  
  0d2ccb8d
- Merge branch 'v0.7.2-dev' into v0.7.2-pa · 6a72a6b4
  zhuwenwen authored Mar 15, 2025
  
  6a72a6b4
- Merge branch 'v0.7.2-dev_yangql' into 'v0.7.2-dev' · 87351a28
  zhuwenwen authored Mar 15, 2025
```
V0.7.2 dev yangql

See merge request dcutoolkit/deeplearing/vllm!85
```
  87351a28
- 支持fusemoe对int4的scale合zero合并读取操作 · 50424a23
  yangql authored Mar 15, 2025
  
  50424a23
- 支持fusemoe对int4的scale合zero合并读取操作 · f78a0510
  yangql authored Mar 15, 2025
  
  f78a0510
- Merge remote-tracking branch 'origin/0.7.2-zhangshao' into v0.7.2-pa · 87bdb89f
  zhuwenwen authored Mar 15, 2025
  
  87bdb89f
- Merge branch 'v0.7.2-pa' of http://10.6.10.68/dcutoolkit/deeplearing/vllm into v0.7.2-pa · e6fd8fda
  zhuwenwen authored Mar 15, 2025
  
  e6fd8fda
14 Mar, 2025 5 commits
- Merge branch 'v0.7.2-dev-quant' into 'v0.7.2-dev' · 9c7191c3
  zhuwenwen authored Mar 14, 2025
```
V0.7.2 dev deepseek v3/r1 block-int8量化支持

See merge request dcutoolkit/deeplearing/vllm!83
```
  9c7191c3
- 解决bf16精度问题，解决cudagraph精度问题 · 4f8d38c8
  zhangshao authored Mar 14, 2025
  
  4f8d38c8
- 增加线性int8 gemm配置 · 268d8a77
  gaoqiong authored Mar 14, 2025
  
  268d8a77
- [feat]add gen_mla_pa_tables.py and optimize triton config configuration · 71c60bd5
  zhuwenwen authored Mar 14, 2025
  
  71c60bd5
- 增加deepseek v3 block-int8量化支持 · 92545504
  gaoqiong authored Mar 14, 2025
  
  92545504
13 Mar, 2025 3 commits
- 优化bf16精度 · b3992aad
  zhangshao authored Mar 13, 2025
  
  b3992aad
- Merge remote-tracking branch 'origin/0.7.2-zhangshao' into v0.7.2-pa · fbde1e5a
  zhuwenwen authored Mar 13, 2025
  
  fbde1e5a
- update mla to obtain the optimal configuration from config · 146eb9d3
  zhuwenwen authored Mar 13, 2025
  
  146eb9d3
12 Mar, 2025 1 commit

Merge branch 'v0.7.2-dev-wm' into 'v0.7.2-dev' · c1370857

zhuwenwen authored Mar 12, 2025

[feat]添加BW deepseek_v3 fused_moe configs

See merge request dcutoolkit/deeplearing/vllm!80

c1370857