configs/mr-potato-head.yaml

pretrained_model_path: "./checkpoints/mr-potato-head"
output_dir: "./outputs/mr-potato-head_lr3e-5_seed33"

train_data:
  video_path: "data/man-surfing.mp4"
  prompt: "a man is surfing"
  n_sample_frames: 8
  width: 512
  height: 512
  sample_start_idx: 0
  sample_frame_rate: 1

validation_data:
  prompts:
    - "sks mr potato head is surfing"
    - "sks mr potato head, wearing a pink hat, is surfing"
    - "sks mr potato head, wearing funny sunglasses, is surfing"
    - "sks mr potato head is surfing in the forest"
  video_length: 8
  width: 512
  height: 512
  num_inference_steps: 50
  guidance_scale: 7.5

learning_rate: 3e-5
train_batch_size: 1
max_train_steps: 500
checkpointing_steps: 1000
validation_steps: 100
trainable_modules:
  - "attn1.to_q"
  - "attn2.to_q"
  - "attn_temp"

seed: 33
mixed_precision: fp16
use_8bit_adam: False
gradient_checkpointing: True
enable_xformers_memory_efficient_attention: True