nodes/aigc-apps-VideoX-Fun
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
195 Commits 6 Branches 0 Tags
f00a7d480e5df0ffa8acd2f05e5d80d175c6104d
Commit Graph
58 Commits
Author SHA1 Message Date
bubbliiiing f00a7d480e Update one-forcing and causal forcing params 2026-10-30 16:33:15 +08:00
bubbliiiing a15b293128 Update ema and update taehv 2026-10-29 10:59:15 +08:00
Bubbliiiing e8c314fbfa Update Sparkwan, Reward Model Training, FastVideo (#525) 2026-10-28 11:34:50 +08:00
li-lizhe 43681aebea fix(offload): resolve the group offloading transfer stream from the onload device (#528) 
The vendored group-offloading helper (see the "Modified from .../diffusers/v0.36.0"
header) resolved its data-transfer stream by probing for CUDA/XPU only, so
`gpu_memory_mode="model_group_offload"`, which calls `apply_group_offloading(..., use_stream=True)`,
raised on a host whose accelerator is neither of those:

    ValueError: Using streams for data transfer requires a CUDA device, or an Intel XPU device.

even when the accelerator's backend implements streams (for example Ascend NPU, where
`torch.cuda.is_available()` is False while `torch.npu` is available and provides `Stream`).

Resolve the stream from the onload device's own backend instead, and keep rejecting devices
whose backend cannot stream, with a message that names the device type. This is the same
policy upstream diffusers applies in the function this file is derived from.

Signed-off-by: li-lizhe <147392333@qq.com>
 2026-10-27 23:15:38 +08:00
Bubbliiiing 4b7b6402a1 Upgrade GPU Memory Mode architecture (#521) 2026-09-29 19:18:17 +08:00
Bubbliiiing 67e3b4b818 Update Qwen Image 2.1 Control and Flex Forcing (#519) 2026-09-24 15:52:16 +08:00
Bubbliiiing 2beb171099 Support Qwen-Image2.1 && Support new Control for Minimax-H3 && Support Wan2.2 with 2.2VAE && Update new tqdm for All Training Code (#511) 2026-09-22 20:41:39 +08:00
hkz 968f0e2192 Add PDD for MiniMax-H3 (#515) 2026-09-04 17:02:21 +08:00
Bubbliiiing 6f3fb60dad Update scale fp8 with fsdp (#508) 2026-08-25 12:35:36 +08:00
Bubbliiiing b0acf916c2 Update Lingbot Worl, Lingbot Video, Minimax-H3 and Minimax-H3 Control (#506) 2026-08-24 16:47:38 +08:00
hkz 248ab0ac0e Add Causal-Focing for Wan2.1 (#500) 2026-07-22 17:37:05 +08:00
Bubbliiiing 403f1f7b78 Fix Bug in Self Forcing in Multi-Gpus Infernece && Update Training Code and Docs && Update Reward Models (#498) 2026-07-14 10:30:54 +08:00
Bubbliiiing 1fd9ed9208 Ode training && Update Lens model && Update LTX2 upsampler (#497) 2026-06-09 15:20:04 +08:00
Bubbliiiing 34036517a8 Added LTX-2.3 support, optimized memory usage by casting e and e0 types in Wan-based models, and updated READMEs for image training and digital human models. (#483) 2026-04-16 20:45:16 +08:00
Bubbliiiing 54bf97ab66 Fix bug in FA3 and LTX2 && Update FA4 support && Update Infinitalk && Update MOVA && Update FlashHead (#479) 2026-04-07 14:08:39 +08:00
Bubbliiiing 4a86483cc2 Reformat S2V models && Update LTX-2 (#476) 2026-03-20 10:45:57 +08:00
Bubbliiiing ad72867c0f Fix Bug in Group Offload when diffusers version is low. (#474) 2026-03-10 15:02:16 +08:00
Bubbliiiing 7a085c9535 Update Longcat Avatar (#453) 2026-02-13 10:14:14 +08:00
Bubbliiiing 1ca4162447 Support validations in all models && Update Saving Code && Preparing Code (#452) 2026-02-10 17:23:23 +08:00
Bubbliiiing ee44fbc950 Lora Loading Log and Z Image Distill (#450) 2026-02-04 16:55:24 +08:00
Bubbliiiing a6b026526f Update Qwen Image Layered, Turbowan Training Code and Z Image predict Code (#448) 2026-02-03 18:23:45 +08:00
Bubbliiiing 95737c8f1d Update Qwen Image Comfyui and pyproject.toml (#437) 2026-01-20 21:03:57 +08:00
Bubbliiiing b534ebfef1 Update Z Image Control Distill Training Code and Update LogValidation in QwenImage and ZImage (#435) 2026-01-16 15:05:49 +08:00
Bubbliiiing 2faa27e6b0 Update LongCatVideo (#419) 2025-12-30 10:11:19 +08:00
Bubbliiiing df665a82c8 Fix bug in Z Image and Qwen Image prediction && Fix the bug in using peft lora in comfyui && Support cfg in Z Image Control (#415) 2025-12-22 15:59:27 +08:00
Bubbliiiing 6a95acc2b7 Flux2 control, Z-Image and Z-Image Control (#384) 2025-12-02 14:28:19 +08:00
Bubbliiiing 200e1f3224 Update flux2 (#383) 2025-11-28 10:31:06 +08:00
Bubbliiiing 599e797e9e Update HunyuanVideo model and Wan2.2 5B Readme (#380) 2025-11-25 13:38:15 +08:00
Bubbliiiing 037a2e8360 Update Peft Lora && Update Readme (#376) 2025-11-20 17:44:22 +08:00
Bubbliiiing 7fd9594919 Fix bug in wan-animate and fantasytalking multi gpus inference && Update Training Codes && Fix bug in s2v lora merging && Update qwen image quick loading (#368) 2025-11-11 14:16:41 +08:00
Bubbliiiing df77df019e Wan animate inference and training (#365) 2025-11-04 13:36:51 +08:00
Bubbliiiing 6183d41df8 Update unload model comfyui for saving VRAM. (#340) 2025-09-28 14:45:22 +08:00
Bubbliiiing e1c1a9562f Update qwenimage edit && Qwenimage comfyuii && Update empty cache before saving (#334) 2025-09-24 11:15:39 +08:00
Bubbliiiing 3e8d6867d3 Vace training && Comfyui update && Fix flux readme && Update MV2V demo (#328) 2025-09-18 15:52:03 +08:00
Bubbliiiing a40638f853 Update lora load in comfyui && Fix bug in qwen image training. (#318) 2025-09-15 13:45:54 +08:00
Bubbliiiing abeac38889 Update Wan2.2 Vace Fun (#316) 2025-09-12 14:44:25 +08:00
Bubbliiiing 74426601fa Update import (#291) 2025-09-08 11:43:02 +08:00
Bubbliiiing b9d3b2aacd Update Flux && Fix bug in text encoder loading && Remove useless import && Fix bug in fun models training (#304) 2025-09-02 14:55:29 +08:00
Bubbliiiing 2e165b7216 Update qwen-image && Fix bug in motion_sub_loss and subject_ref (#300) 
* Update qwen-image
 2025-09-01 13:53:56 +08:00
Bubbliiiing 9c05f3a741 Fix bug in Wan2.2 (#265) 2025-07-29 20:22:28 +08:00
Bubbliiiing 2d3a5cc7d6 Update Wan2.2 (#264) 2025-07-29 00:06:55 +08:00
Bubbliiiing 9436ff2f6e update discrete_sampler (#249) 2025-07-09 13:49:28 +08:00
liubo0902and宣源 c2d0c7ae21 add ac (#225) 
Co-authored-by: 宣源 <xuanyuan.lb@alibaba-inc.com>
 2025-06-26 16:17:03 +08:00
Bubbliiiing 86028a9ce6 Update fix sample size && Fix vram memory bug in wan2.1 lora training (#227) 2025-06-24 10:09:23 +08:00
Bubbliiiing ba1da31fb5 Merge pull request #224 from aigc-apps/support_load_wan_fun_lora 
Support loading Wan-Fun LoRAs into Wan base models
 2025-06-19 16:17:07 +08:00
bubbliiiing e4c74a12f2 Update phantom && add fsdp training && Fix bug in image encoder while using sage attention 2025-06-09 08:25:49 +00:00
huangkunzhe.hkz fab50bd721 Support loading Wan-Fun LoRAs 2025-06-09 11:17:22 +08:00
Bubbliiiing 1061378b66 Add bfloat16 norm && fix bug in cfg_optimization (#217) 2025-05-29 11:28:00 +08:00
Bubbliiiing a1e6ea6335 Pai sparse attention (#211) 
* pai fuser sparse test

* make pai fuser more clear

* make pai fuser more clear

* make pai fuser more clear

* make pai fuser more clear

* Update Readme

* disable compile in rope for less error info

* Fix sage attention backward bug

* Fix checkpoint bugs

* Fix Attention

* Update predict and fast rope

* Fix bug in Sparse

* Fix bug in Sparse

* Fix bug in loras load

* Delete useless import

* Fix bug in ui

* Fix bug in ui
 2025-05-26 16:00:58 +08:00
Bubbliiiing ed4f10beef Fix bug in lora training && Update zero3 cpu offload && Teacache thresholds (#209) 2025-05-13 20:54:20 +08:00