wan (14b T2V)

Description:

Trained only on videos.

• dataset: 135 videos 768x480x41
• steps: 17000
• LR: 2e-4
• optimizer: adamw
• scheduler: contant
• rank: 128
• batch size: 1
• gradient accumulation steps: 1

For training I used the official LTX-Video Trainer repo.