Commits · 51be3ad2f9d42319a5ba2a3dbb9206709c7b32c3 · xuwx1 / LightX2V

14 Nov, 2025 1 commit
- [Fix] remove d2h of cpu-offload infer (#476) · 51be3ad2
  gushiqiao authored Nov 14, 2025
  
  51be3ad2
13 Nov, 2025 1 commit
- [feat]: support matrix game2 universal, gta_drive, templerun & streaming mode · 9826b8ca
  Watebear authored Nov 13, 2025
  
  9826b8ca
07 Nov, 2025 1 commit
- update 5090 int8 config (#449) · e39d8438
  gushiqiao authored Nov 07, 2025
  
  e39d8438
03 Nov, 2025 1 commit

[Fix] fix bias is none (#430) · f0f13701

gushiqiao authored Nov 03, 2025


Co-authored-by: root <root@pt-731f15f5c62740f495519352dc90d9c4-worker-0.pt-731f15f5c62740f495519352dc90d9c4.ns-devsft-3460edd0.svc.cluster.local>

f0f13701

31 Oct, 2025 1 commit
- [Feat] Enable T5 inference and offload overlap for improved efficiency (#423) · 6062ef24
  gushiqiao authored Oct 31, 2025
```
Co-authored-by: gushiqiao <975033167@qq.ocm>
```
  6062ef24
30 Oct, 2025 1 commit
- support lightx2v-kernel and update convert.py (#413) · e106ff67
  Bilang ZHANG authored Oct 30, 2025
  
  e106ff67
17 Oct, 2025 1 commit
- [Recon] remove mm_config and support load single safetensors file (#375) · 94e593ce
  gushiqiao authored Oct 17, 2025
  
  94e593ce
24 Sep, 2025 1 commit
- [Fix] Fix moe offload bug (#330) · 7a111e37
  gushiqiao authored Sep 24, 2025
  
  7a111e37
18 Sep, 2025 1 commit
- [Fix] Fix distribute load model bug (#315) · 3b896f9c
  gushiqiao authored Sep 18, 2025
  
  3b896f9c
16 Sep, 2025 2 commits
- [Fix] Fix high peak memory bug (#313) · f085ede3
  gushiqiao authored Sep 16, 2025
  
  f085ede3
- [Fix] Fix oom bug (#307) · bd21f14f
  gushiqiao authored Sep 16, 2025
  
  bd21f14f
25 Aug, 2025 1 commit
- [Feat] Support vace offload and recon offload. (#245) · abeb9bc8
  gushiqiao authored Aug 25, 2025
  
  abeb9bc8
06 Aug, 2025 3 commits
- Support loading bf16 weights and converting them to fp32 · 94bd4599
  gushiqiao authored Aug 06, 2025
  
  94bd4599
- Support loading bf16 weights and converting them to fp32 · 6b32b743
  gushiqiao authored Aug 06, 2025
  
  6b32b743
- Support bf16/fp16 inference and mixed-precision inference with fp32 for some layers · 3e215bad
  gushiqiao authored Aug 06, 2025
  
  3e215bad
05 Aug, 2025 3 commits
- Support wan2.2-distill models (#186) · a7d70484
  Zhuguanyu Wu authored Aug 05, 2025
```
* add gguf loading functions

* support wan22 distill
```
  a7d70484
- style: remove unused import (#184) · c0b36010
  PengGao authored Aug 05, 2025
  
  c0b36010
- style: add ruff isort (#183) · f6d4fc85
  PengGao authored Aug 05, 2025
  
  f6d4fc85
21 Jul, 2025 3 commits
- wrong fix · 9b2c5b6b
  Wq-dd authored Jul 21, 2025
  
  9b2c5b6b
- fix import torch.ao · a6dff8f4
  Wq-dd authored Jul 21, 2025
  
  a6dff8f4
- Add torchao kernel · 7f2fbd00
  gushiqiao authored Jul 21, 2025
  
  7f2fbd00
12 Jul, 2025 1 commit
- FIX · 1b5d336d
  gushiqiao authored Jul 12, 2025
  
  1b5d336d
16 Jun, 2025 2 commits
- Fix · a92ea6e8
  gushiqiao authored Jun 16, 2025
  
  a92ea6e8
- Fixed the accuracy fluctuation bug · 8ba6e3b4
  gushiqiao authored Jun 16, 2025
  
  8ba6e3b4
11 Jun, 2025 1 commit
- Support gradio demo for wan i2v. · 61420957
  gushiqiao authored Jun 11, 2025
  
  61420957
10 Jun, 2025 1 commit
- Update cache config and fix bugs. · 7ec21ca6
  gushiqiao authored Jun 10, 2025
  
  7ec21ca6
09 Jun, 2025 2 commits

fix. · 9121bad1
gushiqiao authored Jun 09, 2025

9121bad1

Support run load memory machine, fix some bugs and reconstruct quantizaton. (#61) · 5c241f86

gushiqiao authored Jun 09, 2025



* reconstruct quantization and fix memory leak bug.

* Support lazy load inference.

* reconstruct quantization

* Fix hunyuan bugs

* deleted tmp file

---------
Co-authored-by: root <root@pt-c0b333b3a1834e81a0d4d5f412c6ffa1-worker-0.pt-c0b333b3a1834e81a0d4d5f412c6ffa1.ns-devsft-3460edd0.svc.cluster.local>
Co-authored-by: gushiqiao <gushqiaio@sensetime.com>
Co-authored-by: gushiqiao <gushiqiao@sensetime.com>

5c241f86

28 May, 2025 1 commit

Xinchi/fix offload (#57) · 2bb1b0f0

Xinchi Huang authored May 28, 2025



* fix offload extra latency in the first step by pre-allocating pinned memory

* pre-commit

---------
Co-authored-by: “de1star” <“843414674@qq.com”>

2bb1b0f0

14 May, 2025 1 commit

fix offload & peak memory decorator · e74270f5

Xinchi Huang authored May 14, 2025



* fix offload

* fix offload

---------
Co-authored-by: “de1star” <“843414674@qq.com”>

e74270f5

09 May, 2025 2 commits

Support load advance ptq model. (#33) · 165ec807

gushiqiao authored May 09, 2025



* Support load advance ptq model.

* Update run_wan_i2v_advanced_ptq.sh

---------
Co-authored-by: gushiqiao <gushiqiao@sensetime.com>
Co-authored-by: Yang Yong(雍洋) <yongyang1030@163.com>

165ec807

update save/load of mm weights · 53c0d05c
helloyongyang authored May 09, 2025

53c0d05c

07 May, 2025 2 commits
- Fix weight loading without quant · dbfa688b
  helloyongyang authored May 07, 2025
  
  dbfa688b
- Support save_naive_quant and load quantization weight · 420fec7f
  helloyongyang authored May 07, 2025
  
  420fec7f
29 Apr, 2025 1 commit
- Support convert weight to diffusers. · 6347b21b
  root authored Apr 29, 2025
  
  6347b21b
28 Apr, 2025 1 commit
- update some variable names · aec90a0d
  helloyongyang authored Apr 28, 2025
  
  aec90a0d
11 Apr, 2025 1 commit

Support hunyuan offload and teacache. (#15) · 4eec372d

gushiqiao authored Apr 11, 2025



* Support hunyuan offload and teacache.

* Fix

* Fix

---------
Co-authored-by: gushiqiao <gushiqiao@sensetime.com>

4eec372d

09 Apr, 2025 2 commits
- fix bug & add quant kernel · 6c18f54c
  helloyongyang authored Apr 09, 2025
  
  6c18f54c
- [Feature]: support many quant kernels · 3aa95081
  helloyongyang authored Apr 09, 2025
  
  3aa95081
08 Apr, 2025 1 commit
- Support sync cpu offload. (#10) · 683aaa3a
  gushiqiao authored Apr 04, 2025
```
Co-authored-by: gushiqiao <gushiqiao@sensetime.com>
```
  683aaa3a