5 Pull requests merged by 4 people

• Update wan_video.py

  #493 merged Mar 26, 2025

• [example] add usp (unified sequence parallel) for wanx example

  #479 merged Mar 25, 2025

• support Unified Sequence Parallel

  #485 merged Mar 25, 2025

• Update README.md

  #482 merged Mar 25, 2025

• update tensor parallel

  #480 merged Mar 25, 2025

7 Issues closed by 6 people

• Wan 1.3B T2V requires 48 GB VRAM

  #492 closed Mar 27, 2025

• NCCL timeout when fine-tuing use 8*A40 44G

  #489 closed Mar 26, 2025

• Fine-tuning Wan 1.3B T2V for I2V

  #488 closed Mar 26, 2025

• Fail to add lora when test lora for 14B I2V

  #478 closed Mar 26, 2025

• Wan-Video-14B-I2V多卡加速

  #487 closed Mar 25, 2025

• Can't load OnomaAIResearch/Illustrious-XL-v1.1

  #472 closed Mar 24, 2025

• is wan-vae cache mechine enabled during training phase?

  #471 closed Mar 21, 2025

16 Issues opened by 15 people

• How to use tensorboard?

  #497 opened Mar 28, 2025

• Does wan2.1 14b support Lora？

  #496 opened Mar 27, 2025

• tensor size error when using wan_14b_text_to_video_tensor_parallel in more than 2 gpus

  #495 opened Mar 27, 2025

• We cannot detect the model type. No models are loaded.

  #494 opened Mar 27, 2025

• WanX 14B I2V "Cannot load LoRA"

  #491 opened Mar 26, 2025

• When I used Lora to train Wan2.1-1.3b-T2V, the result was snowflakes

  #490 opened Mar 26, 2025

• OOM: When the input video resolution is 720P, LoRA training will run out of memory.

  #486 opened Mar 25, 2025

• Sage Attention vs Flash Attention Speed Comparison with Wan 2.1 - Sage Attention is 37% Faster - 720p - 14b model - tested on Windows Python VENV - no WSL

  #484 opened Mar 25, 2025

• Wan2.1 issue

  #483 opened Mar 25, 2025

• Possible Support on Stepvideo-ti2v

  #481 opened Mar 25, 2025

• flash_attn._flash_attn_forward.default does not have a sharding strategy registered

  #477 opened Mar 24, 2025

• Please support GGUF models for Wan 2.1 - currently 16 GB VRAM GPUs cant run 720p they can with GGUF

  #476 opened Mar 24, 2025

• Compararison with diffusers

  #475 opened Mar 24, 2025

• inconsistency with the official implementation of WanX2.1

  #474 opened Mar 24, 2025

• Issues regarding the compatibility of flux lora

  #473 opened Mar 24, 2025

• Hunyuan I2V model fails

  #470 opened Mar 21, 2025

11 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• stable diffusion sdxl train error

  #253 commented on Mar 24, 2025 • 0 new comments

• 推理wan2.1图生视频模型如何开启flash attention？

  #469 commented on Mar 24, 2025 • 0 new comments

• wan2.1-i2v-14b微调lora后是否可以加载fp8的lora，目前lora应该是fp16？或者有什么加速的方法吗？目前生成360p的5s视频需要20分钟。

  #467 commented on Mar 24, 2025 • 0 new comments

• Trying to run wan_14b_text_to_video.py example on a 12G card, tries to allocate 160 GiB of VRAM and fails

  #462 commented on Mar 24, 2025 • 0 new comments

• Multi-node Multi-gpu Wan1.3b training

  #464 commented on Mar 24, 2025 • 0 new comments

• [Bug] Fix flash_attn_func Return Value Handling for flash-attn3 Compatibility in wan_video_dit Model

  #468 commented on Mar 24, 2025 • 0 new comments

• Wan 2.1 FP8 model weights causing color issue - BF16 has no issue

  #466 commented on Mar 24, 2025 • 0 new comments

• support 14B full training？

  #422 commented on Mar 25, 2025 • 0 new comments

• NUS-HPC-AI-Lab Enhance-A-Video for Wan 2.1 arrived can we get it?

  #428 commented on Mar 25, 2025 • 0 new comments

• Any chance to implement RIFLEx ?

  #377 commented on Mar 26, 2025 • 0 new comments

• This pull request is for wan_video.py to add cancel feature

  #463 commented on Mar 26, 2025 • 0 new comments