Yishuo Wang
6249c1e373
rewrite llama optimization ( #12609 )
2024-12-25 17:04:32 +08:00
Yishuo Wang
5f5ac8a856
fix llama related import ( #12611 )
2024-12-25 16:23:52 +08:00
Yishuo Wang
4e6b9d804f
add compresskv back for mistral ( #12607 )
...
* add compresskv back for mistral
* fix
* fix
2024-12-25 11:06:08 +08:00
Yishuo Wang
4135b895b3
refactor chatglm2, internlm, stablelm and qwen ( #12604 )
2024-12-24 18:18:00 +08:00
Yishuo Wang
073f936c37
refactor mistral and phi3 ( #12605 )
2024-12-24 17:52:32 +08:00
Yishuo Wang
ad2dc965c5
refactor mllama, gpt2 and internvl ( #12602 )
2024-12-24 14:18:31 +08:00
Yishuo Wang
7aaf02f602
refactor baichuan, glm4 and minicpm3 ( #12600 )
2024-12-24 14:16:30 +08:00
Yishuo Wang
098eb335b2
refactor sd 1.5 and qwen2-vl and fix ( #12590 )
2024-12-20 17:34:55 +08:00
Yishuo Wang
b050368efc
refactor yuan2 and starcoder2 and fix ( #12589 )
2024-12-20 16:41:50 +08:00
Yishuo Wang
6ea8033635
refactor glm edge ( #12588 )
2024-12-20 15:36:57 +08:00
Yishuo Wang
f3b5fad3be
refactor qwen2 and llama3 ( #12587 )
2024-12-20 13:25:25 +08:00
Yishuo Wang
3eeb02f1be
support Megrez-3B-Omni ( #12582 )
2024-12-19 17:23:01 +08:00
Yishuo Wang
80f2fdc37b
optimize new minicpm model ( #12579 )
2024-12-19 14:22:47 +08:00
Yishuo Wang
4540424271
optimize siglip attention again ( #12578 )
2024-12-19 13:40:48 +08:00
Yishuo Wang
e0921f80c1
padding mask on torch side ( #12577 )
2024-12-19 10:53:02 +08:00
Yishuo Wang
e2ae42929a
small fix ( #12573 )
2024-12-18 15:48:22 +08:00
Yishuo Wang
a4eb561f36
optimize siglip attention on arc ( #12569 )
2024-12-18 14:19:43 +08:00
Yishuo Wang
a608f26cc8
use new fused layer norm ( #12553 )
2024-12-17 13:52:35 +08:00
Yishuo Wang
5ae0006103
remove old rope usage ( #12552 )
2024-12-16 15:59:36 +08:00
Yishuo Wang
c090d167dc
remove old rope usage ( #12544 )
2024-12-13 16:54:58 +08:00
Yishuo Wang
15219944b8
optimize glm edge again ( #12539 )
2024-12-13 13:52:39 +08:00
Yishuo Wang
ffce86d69f
add basic glm-edge-v support ( #12533 )
2024-12-12 17:25:48 +08:00
Yishuo Wang
3e0823d2ae
add basic glm-edge support ( #12531 )
2024-12-12 16:02:22 +08:00
Yishuo Wang
77404d2a63
support new model ( #12523 )
2024-12-11 13:41:15 +08:00
Yishuo Wang
a9e3f7f14c
optimize minicpm ( #12496 )
2024-12-04 17:14:16 +08:00
Yishuo Wang
e0bf0054e1
small fix ( #12493 )
2024-12-04 16:37:39 +08:00
Yishuo Wang
5629fdd518
optimize qwen2_vl multiple image input or video input ( #12487 )
2024-12-04 09:24:38 +08:00
Yishuo Wang
6f3441ba4c
fix glm4-9b overflow ( #12455 )
2024-11-27 17:39:13 +08:00
Jinhe
66bd7abae4
add sdxl and lora-lcm optimization ( #12444 )
...
* add sdxl and lora-lcm optimization
* fix openjourney speed drop
2024-11-26 11:38:09 +08:00
Yishuo Wang
cdd41f5e4c
optimize sdxl again ( #12441 )
2024-11-25 17:46:46 +08:00
Yishuo Wang
8164aed802
small change ( #12439 )
2024-11-25 14:35:49 +08:00
Yishuo Wang
be132c4209
fix and optimize sd ( #12436 )
2024-11-25 14:09:48 +08:00
Yuwen Hu
8fdc36c140
Optimize with new batch kernel when batch_size=1 on LNL ( #12419 )
...
* Add use batch kernel condition for LNL
* Fix for other device judgement
* Fix based on comment
2024-11-21 16:21:35 +08:00
Yuwen Hu
a69395f31f
Support performance mode of GLM4 model ( #12401 )
...
* Initial support of prepare generation args for transformers 445
* Small fix to chatglm4 model optimization
* Small fix
* fix glm4 position id
* fix glm4 error
* Small change in conditon & fix based on comments
* Style fixes
---------
Co-authored-by: cyita <yitastudy@gmail.com>
2024-11-18 18:46:52 +08:00
Yishuo Wang
dc34e8c51f
optimize glm4v vision attention ( #12369 )
2024-11-08 17:01:57 +08:00
Yuwen Hu
1a6cbc473f
Add fused mlp optimizations to glm4 models ( #12360 )
...
* Add fused mlp to glm4 models
* Small fix
2024-11-07 18:52:47 +08:00
Yishuo Wang
ad68c56573
small improvement ( #12359 )
2024-11-07 15:57:41 +08:00
Yuwen Hu
872a74481a
Small optimization to glm4 models ( #12351 )
2024-11-06 19:16:58 +08:00
Yina Chen
f24352aef9
llama 3.1/3.2 support compresskv ( #12347 )
...
* llama 3.1/3.2 support compresskv
* update
* fix transformers 4.45 error
* fix style
* fix typo
* disable llama3.2 1b compresskv
2024-11-06 17:33:43 +08:00
Yishuo Wang
e23ef7d088
optimize glm4v's vision part ( #12346 )
2024-11-06 15:43:40 +08:00
Zhao Changmin
1b637e4477
Add chatglm2&3 fuse mlp ( #12328 )
...
* add chatglm fuse mlp
2024-11-04 18:04:41 +08:00
Yishuo Wang
b9853f98b3
fix qwen2 attention_mask slice ( #12307 )
2024-10-31 17:00:05 +08:00
Xin Qiu
97a0f7fd35
Codegeex support ( #12303 )
...
* new codegeex attn
* use kv cache
* add compress/quantize kv
* remove compress/quantize kv
* fix style check
* fix style
* fix codegeex
2024-10-31 15:28:56 +08:00
Yishuo Wang
72605c7016
fix llama3.1/3.2 quantize kv check ( #12302 )
2024-10-31 11:55:07 +08:00
Yishuo Wang
540eaeb12c
refactor attention_softmax ( #12295 )
2024-10-30 13:20:50 +08:00
Xin Qiu
39c9d1de52
fix code geex ( #12261 )
2024-10-24 14:34:01 +08:00
Yishuo Wang
f3a2b20e6b
Optimize gpt2 ( #12259 )
2024-10-24 13:44:24 +08:00
Yishuo Wang
9ea694484d
refactor ot remove old rope usage ( #12224 )
2024-10-17 17:06:09 +08:00
Yishuo Wang
324bcb057e
refactor to reduce old rope usage ( #12219 )
2024-10-17 14:45:09 +08:00
Yishuo Wang
a4a758656a
refactor gemma to reduce old fuse rope usage ( #12215 )
2024-10-16 17:40:28 +08:00
