Commits · 9d5187eb68ef31f5fb5d63f4e3b4862fb186aad2 · OpenDAS / vllm_cscc · GitLab

13 Jun, 2025 1 commit
- 解耦w8a8 的gemm以及moe算子相关实现 · 942368c7
  gaoqiong authored Jun 13, 2025
  
  942368c7
05 Jun, 2025 2 commits
- 新增qwen3 fusemoe优化支持 · 1cb37dab
  yangql authored Jun 05, 2025
  
  1cb37dab
- 修改deepseelk block-int8 权重处理流程，增加per-channel bestconfig配置以及首次triton warmup代码 · 40b94473
  gaoqiong authored Jun 05, 2025
  
  40b94473
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
27 May, 2025 2 commits
- [feat]适配sgl moe_fused_gate kernel · 0b229519
  王敏 authored May 27, 2025
  
  0b229519
- 修复0.8.5 config找不到的bug · 823b3332
  gaoqiong authored May 27, 2025
  
  823b3332
23 May, 2025 1 commit
- tbo增加ds int8支持，修改tb zerooverhead环境变量到envs.py中统一管理 · dbbb148b
  lizhigong authored May 23, 2025
  
  dbbb148b
22 May, 2025 2 commits
- 优化awq权重转换的方式，修复可能会导致蒸馏模型加载时报错的bug · a94ed3ea
  yangql authored May 22, 2025
  
  a94ed3ea
- fix tbo to support deepseek · 0cc7c880
  lizhigong authored May 22, 2025
  
  0cc7c880
15 May, 2025 1 commit
- Update deprecated type hinting in `models` (#18132) · 26d04193
  Harry Mellor authored May 15, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  26d04193
14 May, 2025 1 commit
- Modularize fused experts and integrate PPLX kernels (#15956) · f9c069c8
  bnellnm authored May 14, 2025
  
  f9c069c8
09 May, 2025 1 commit
- [Attention] MLA move rotary embedding to cuda-graph region (#17668) · 5e6f9394
  Lucas Wilkinson authored May 08, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  5e6f9394
02 May, 2025 1 commit
- [Attention] MLA move o_proj q_proj into cuda-graph region (#17484) · afcb3f88
  Lucas Wilkinson authored May 01, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
```
  afcb3f88
24 Apr, 2025 1 commit
- [Chore] Remove Sampler from Model Code (#17084) · b411418f
  Woosuk Kwon authored Apr 24, 2025
```
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
```
  b411418f
16 Apr, 2025 1 commit
- update moe models of LM_NN · 2db90e2c
  zhuwenwen authored Apr 16, 2025
  
  2db90e2c
11 Apr, 2025 1 commit
- Revert "[Model] use AutoWeightsLoader for deepseek_v2, internlm2" (#16453) · 905e91e9
  DefTruth authored Apr 11, 2025
  
  905e91e9
10 Apr, 2025 1 commit
- [Model] use AutoWeightsLoader for deepseek_v2, internlm2 (#16383) · a9bd832f
  Aaron Ang authored Apr 10, 2025
```
Signed-off-by: Aaron Ang <aaron.angyd@gmail.com>
```
  a9bd832f
08 Apr, 2025 1 commit

[Bugfix] fix deepseek fp16 scale bug (#14809) · db104221

Jinzhen Lin authored Apr 09, 2025


Signed-off-by: Jinzhen Lin <linjinzhen@hotmail.com>
Co-authored-by: mgoin <mgoin64@gmail.com>

db104221

07 Apr, 2025 1 commit
- 增加fused-moe int4/int8的支持，以及deepseek精度问题的修复 · bb94d2e5
  yangql authored Apr 07, 2025
  
  bb94d2e5
01 Apr, 2025 1 commit
- 去掉自研专家并行代码 · 9beadbfb
  王敏 authored Apr 01, 2025
  
  9beadbfb
20 Mar, 2025 1 commit
- [Frontend][Bugfix] support prefill decode disaggregation on deepseek (#14824) · 742369d3
  billishyahao authored Mar 20, 2025
```
Signed-off-by: billishyahao <bill.he@amd.com>
Co-authored-by: Zhai Feiyue <80079571+ZhaiFeiyue@users.noreply.github.com>
```
  742369d3
15 Mar, 2025 1 commit
- 支持fusemoe对int4的scale合zero合并读取操作 · 2b47d4fa
  zhuwenwen authored Mar 15, 2025
  
  2b47d4fa
14 Mar, 2025 1 commit
- 增加deepseek v3 block-int8量化支持 · ba1ff372
  zhuwenwen authored Mar 14, 2025
```
增加线性int8 gemm配置
```
  ba1ff372
11 Mar, 2025 1 commit
- [Bugfix] Fix FP16 overflow for DeepSeek V2 (#13232) · c982ac57
  Concurrensee authored Mar 10, 2025
```
Signed-off-by: Yida Wu <yida.wu@amd.com>
```
  c982ac57
04 Mar, 2025 1 commit
- Clean up unused padding_idx variables across many model definitions (#13240) · 4f5b059f
  Tyler Michael Smith authored Mar 04, 2025
```
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
```
  4f5b059f
03 Mar, 2025 2 commits
- [Bugfix] Allow shared_experts skip quantization for DeepSeekV2/V3 (#14100) · 2b04c209
  Michael Goin authored Mar 03, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  2b04c209
- [Misc] duplicate code in deepseek_v2 (#14106) · e584b85a
  wang.yuqi authored Mar 03, 2025
  
  e584b85a
27 Feb, 2025 2 commits
- [Attention] MLA support for V1 (#13789) · 58d1b2aa
  Yang Chen authored Feb 27, 2025
```
Signed-off-by: Yang Chen <yangche@fb.com>
```
  58d1b2aa
- update mtp layout · c5888d31
  zhuwenwen authored Feb 27, 2025
  
  c5888d31
26 Feb, 2025 2 commits
- update deepseek-v2 · 96802ca0
  zhuwenwen authored Feb 26, 2025
  
  96802ca0
- DeepSeek V2/V3/R1 only place `lm_head` on last pp rank (#13833) · 24679788
  Harry Mellor authored Feb 26, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  24679788
25 Feb, 2025 2 commits
- [Bugfix] Fix deepseek-v2 error: "missing 1 required positional argument: 'residual'" (#13802) · 4a8cfc75
  Lucas Wilkinson authored Feb 24, 2025
  
  4a8cfc75
- Remove unused kwargs from model definitions (#13555) · cdc1fa12
  Harry Mellor authored Feb 25, 2025
  
  cdc1fa12
24 Feb, 2025 1 commit
- Expert Parallelism (EP) Support for DeepSeek V2 (#12583) · 781096e3
  Jongseok Park authored Feb 24, 2025
  
  781096e3
20 Feb, 2025 1 commit
- 适配deepseekv3\v2 moe awq的推理支持 · 47bd229c
  yangql authored Feb 20, 2025
  
  47bd229c
19 Feb, 2025 2 commits
- [Model][Speculative Decoding] DeepSeek MTP spec decode (#12755) · f525c0be
  Lucia Fang authored Feb 19, 2025
```
Signed-off-by: Lu Fang <fanglu@fb.com>
Co-authored-by: LiuXiaoxuanPKU <lilyliupku@gmail.com>
```
  f525c0be
- update lm_head tn layout for awq · f0e7d72d
  zhuwenwen authored Feb 19, 2025
  
  f0e7d72d
17 Feb, 2025 1 commit
- [fix]1.解决deepseekv2/v3 nn layout加载报错问题;2.上传ep1和ep2的fused_moe config · 249fca2a
  王敏 authored Feb 17, 2025
  
  249fca2a
13 Feb, 2025 1 commit
- support deepseek_v2 nn layout · a8b2d878
  zhuwenwen authored Feb 13, 2025
  
  a8b2d878
12 Feb, 2025 1 commit
- 优化ep moe · 44d451ff
  王敏 authored Feb 12, 2025
  
  44d451ff