-
Hi,
Thanks for your work.
What is the input and output shapes of referencenet model?
I assumed reference_latents and motion_latents should be passed through the referencenet.
Thanks
-
Hi, it seems that you train the 2D unet, referencenet, and poseguider during the **first** stage,
but you don't save parameters of 2D unet.
-
# [WeeklyReports] 2024.03.09~2024.03.22 周报汇总
请各位学员在本 issue 下以 comment 的形式填写周报摘要,ddl 本周五晚,格式示例如下:
```
### 姓名
xxx
### 本周工作
1. xxx
2. xxx
### 下周工作
1. xxx
2. xxx
### 详细周报链接:
…
-
-
-
Great work, and i have some question about the attention modules (spatial attention&cross-attention&temporal attention), but the spatial-attention for calculating reference-net latent feature and deno…
-
-
# ComfyUI Error Report
## Error Details
- **Node Type:** Molmo7BDbnb
- **Exception Type:** ImportError
- **Exception Message:** Failed to load dependencies:
CUDA Setup failed despite G…
-
File "train_th.py", line 637, in
main(name=name, launcher=args.launcher, use_wandb=args.wandb, **config)
File "train_th.py", line 460, in main
latents_pose = poseguider(mask_image)
F…
-
Hi, Thanks for any suggestions.
The largest resolution that could be used for training is 512 × 512 with ~76G memory cost.
I set the enable_xformers_memory_efficient_attention to True but nothing ch…