pretrained_model_path: CompVis/stable-diffusion-v1-4
output_dir: /content/Tune-A-Video/outputs/city
train_data:
  video_path: /content/Tune-A-Video/data/city.mp4
  prompt: wide angle views of a bus with wheels and tires driving through city, a
    city with high detailed street. It's clear sky. It's snowing weather. It's sunrise
  n_sample_frames: 24
  width: 512
  height: 512
  sample_start_idx: 0
  sample_frame_rate: 1
validation_data:
  prompts:
  - wide angle views of a bus with wheels and tires driving through city, a city with
    high detailed street. It's clear sky. It's snowing weather. It's sunrise
  video_length: 24
  width: 512
  height: 512
  num_inference_steps: 50
  guidance_scale: 12.5
  use_inv_latent: true
  num_inv_steps: 50
validation_steps: 1
trainable_modules:
- attn1.to_q
- attn2.to_q
- attn_temp
train_batch_size: 1
max_train_steps: 5
learning_rate: 3.0e-05
scale_lr: false
lr_scheduler: constant
lr_warmup_steps: 0
adam_beta1: 0.9
adam_beta2: 0.999
adam_weight_decay: 0.01
adam_epsilon: 1.0e-08
max_grad_norm: 1.0
gradient_accumulation_steps: 1
gradient_checkpointing: true
checkpointing_steps: 5
resume_from_checkpoint: null
mixed_precision: fp16
use_8bit_adam: false
enable_xformers_memory_efficient_attention: true
seed: 33