Commits · a50eb0e67687298d6a26508769ad4fd543687167 · change / sglang

19 Nov, 2025 3 commits
- Merge remote-tracking branch 'origin/v0.5.4_dev_yiqa' into v0.5.4_dev_yiqa · a50eb0e6
  yiqa authored Nov 19, 2025
  
  a50eb0e6
- Merge branch 'v0.5.4_dev_maxiao' into 'v0.5.4_dev' · 0dc51b09
  maxiao1 authored Nov 19, 2025
```
修复w8a8_marlin tp pp

See merge request OpenDAS/sglang!30
```
  0dc51b09
- 修复w8a8_marlin tp pp · 158efb14
  maxiao1 authored Nov 19, 2025
  
  158efb14
18 Nov, 2025 2 commits
- Merge branch 'v0.5.4_dev_linhai' into 'v0.5.4_dev' · b6e7f2a9
  lizhigong authored Nov 18, 2025
```
V0.5.4 dev linhai

See merge request OpenDAS/sglang!28
```
  b6e7f2a9
- fa3 support qwen. · 4c45697e
  shangxl authored Nov 18, 2025
  
  4c45697e
17 Nov, 2025 2 commits
- Merge branch 'v0.5.4_torch2.7' into 'v0.5.4_dev' · de61a992
  maxiao1 authored Nov 17, 2025
```
sgl_kernel适配torch2.7

See merge request OpenDAS/sglang!26
```
  de61a992
- sgl_kernel适配torch2.7 · 48217ab6
  maxiao authored Nov 17, 2025
  
  48217ab6
15 Nov, 2025 2 commits
- Merge branch 'v0.5.4_dev_maxiao' into 'v0.5.4_dev' · bd63af06
  maxiao1 authored Nov 15, 2025
```
适配w8a8_marlin 高吞吐模式

See merge request OpenDAS/sglang!25
```
  bd63af06
- 适配w8a8_marlin 高吞吐模式 · eed591c9
  maxiao1 authored Nov 15, 2025
  
  eed591c9
14 Nov, 2025 11 commits
- Merge branch 'v0.5.4_dev_yiqa' into 'v0.5.4_dev' · 92f82dce
  lizhigong authored Nov 14, 2025
```
使用groupgemm完成高吞吐模式适配。

See merge request OpenDAS/sglang!24
```
  92f82dce
- 使用groupgemm完成高吞吐模式适配。 · 54abdab4
  yiqa authored Nov 14, 2025
  
  54abdab4
- 使用groupgemm完成高吞吐模式适配 · 474b26ab
  yiqa authored Nov 14, 2025
  
  474b26ab
- 使用groupgemm完成高吞吐模式适配 · 1693e754
  yiqa authored Nov 14, 2025
  
  1693e754
- Merge remote-tracking branch 'origin/v0.5.4_dev_yiqa' into v0.5.4_dev_yiqa · ce363e89
  yiqa authored Nov 14, 2025
```
# Conflicts:
#	python/sglang/srt/layers/quantization/slimquant_w4a8_marlin.py
```
  ce363e89
- 使用groupgemm完成高吞吐模式适配 · 20241efa
  yiqa authored Nov 14, 2025
  
  20241efa
- 使用groupgemm完成高吞吐模式适配 · 8f355853
  yiqa authored Nov 14, 2025
  
  8f355853
- Merge branch 'v0.5.4_dev_maxiao' into 'v0.5.4_dev' · a34b0d3d
  maxiao1 authored Nov 14, 2025
```
适配w8a8_marlin

See merge request OpenDAS/sglang!23
```
  a34b0d3d
- 适配w8a8_marlin · 25796d05
  maxiao1 authored Nov 14, 2025
  
  25796d05
- Merge branch 'v0.5.4_dev_wanghan' into 'v0.5.4_dev' · ac7dcc2d
  maxiao1 authored Nov 14, 2025
```
算子融合

See merge request OpenDAS/sglang!22
```
  ac7dcc2d
- 算子融合 · 3dc7dc6c
  maxiao1 authored Nov 14, 2025
  
  3dc7dc6c
13 Nov, 2025 3 commits
- Merge branch 'v0.5.4_support_compile' into 'v0.5.4_dev' · 842b423a
  lizhigong authored Nov 13, 2025
```
以尽量小的改动支持torch compile，补全分支代码（copy from vllm）

See merge request OpenDAS/sglang!21
```
  842b423a
- 以尽量小的改动支持torch compile，补全部分接口 · c03b2816
  renzhc authored Nov 13, 2025
  
  c03b2816
- Merge branch 'v0.5.4_dev_linhai' into 'v0.5.4_dev' · eb4b015f
  lizhigong authored Nov 13, 2025
```
modify codes with performance issues.

See merge request OpenDAS/sglang!20
```
  eb4b015f
12 Nov, 2025 5 commits
- update to start mtp. · a55cb8b2
  linhai1 authored Nov 12, 2025
  
  a55cb8b2
- Merge branch 'v0.5.4_dev_maxiao' into 'v0.5.4_dev' · cbf7a3e3
  maxiao1 authored Nov 12, 2025
```
使用groupgemm完成低延迟模式适配

See merge request OpenDAS/sglang!19
```
  cbf7a3e3
- 使用groupgemm完成低延迟模式适配 · e7a1f94c
  maxiao1 authored Nov 12, 2025
  
  e7a1f94c
- Merge branch 'v0.5.4_dev' of http://developer.sourcefind.cn/codes/OpenDAS/sglang into v0.5.4_dev · 39f99208
  linhai1 authored Nov 12, 2025
  
  39f99208
- modify codes with performance issues. · f37b05b5
  linhai1 authored Nov 12, 2025
  
  f37b05b5
11 Nov, 2025 6 commits
- Merge branch 'v0.5.4_dev_linhai' into 'v0.5.4_dev' · 3246cea1
  lizhigong authored Nov 11, 2025
```
V0.5.4 dev linhai

See merge request OpenDAS/sglang!17
```
  3246cea1
- support fp8_e4m3 and fp8_e5m2. · 59b01a00
  linhai1 authored Nov 11, 2025
  
  59b01a00
- support fp8_e4m3 and fp8_e5m2. · 34f0ebb1
  linhai1 authored Nov 11, 2025
  
  34f0ebb1
- support fp8_e4m3. · 04ea3540
  linhai1 authored Nov 11, 2025
  
  04ea3540
- support fp8_e4m3. · 50f7ea0f
  linhai1 authored Nov 11, 2025
  
  50f7ea0f
- support fp8_e4m3. · 484c5433
  linhai1 authored Nov 11, 2025
  
  484c5433
10 Nov, 2025 6 commits
- Merge branch 'v0.5.4_dev' into v0.5.4_dev_linhai · 6741925c
  lizhigong authored Nov 10, 2025
  
  6741925c
- Merge branch 'v0.5.4_dev' into v0.5.4_dev_linhai · 12b60933
  lizhigong authored Nov 10, 2025
  
  12b60933
- Merge branch 'v0.5.4_dev_liucong' into 'v0.5.4_dev' · 93eb92f8
  maxiao1 authored Nov 10, 2025
```
V0.5.4 dev liucong

See merge request OpenDAS/sglang!16
```
  93eb92f8
- V0.5.4 dev liucong · 5f643074
  maxiao1 authored Nov 10, 2025
  
  5f643074
- Merge branch 'v0.5.4_dev_lzg' into 'v0.5.4_dev' · 698bc661
  lizhigong authored Nov 10, 2025
```
fix bug on flash attention use for chunkprefill and radix cache

See merge request OpenDAS/sglang!14
```
  698bc661
- Merge branch 'v0.5.4_rzc' into 'v0.5.4_dev' · 1eaad6d1
  lizhigong authored Nov 10, 2025
```
添加环境变量SGLANG_USE_LIGHTOP 控制 lightop的融合rotaty_emb和moe_gated算子，默认禁用；修复RMSNorm:forward_hip中的错误逻辑

See merge request OpenDAS/sglang!13
```
  1eaad6d1