File size: 1,236 Bytes
35069c1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 |
tracker_project_name: memo
output_dir: outputs/finetune
resume_from_checkpoint: null
model_name_or_path: memoavatar/memo
vae: stabilityai/sd-vae-ft-mse
gradient_checkpointing: true
gradient_accumulation_steps: 1
train_batch_size: 1
max_train_steps: 3500
num_train_epochs: -1
enable_xformers_memory_efficient_attention: true
checkpoints_total_limit: 20
robust_training: true
learning_rate: 1.0e-05
max_grad_norm: 1.0
scale_lr: false
lr_scheduler: constant
lr_warmup_steps: 0
seed: 42
mixed_precision: bf16
use_8bit_adam: false
allow_tf32: true
use_ema: false
adam_beta1: 0.9
adam_beta2: 0.999
adam_weight_decay: 0.01
adam_epsilon: 1.0e-08
dataloader_num_workers: 16
prefetch_factor: 4
checkpointing_steps: 5000
data:
width: 512
height: 512
num_past_frames: 16
dynamic_past_frames: false
n_sample_frames: 16
audio_margin: 2
metadata_paths:
- data/embedding/metadata.jsonl
weighting_scheme: logit_normal
logit_mean: 0.0
logit_std: 1.0
mode_scale: 1.29
noise_scheduler_kwargs:
num_train_timesteps: 1000
train_reference_net: false
train_diffusion_net: true
train_image_proj: false
train_audio_proj: false
trainable_modules:
- motion_modules
- audio_modules
uncond_img_ratio: 0.05
uncond_audio_ratio: 0.05
start_ratio: 0.05
|