colab/infer.yaml

# train
resolution_w: 512
resolution_h: 512
video_length: 30
enable_xformers_memory_efficient_attention: True
seed: 12580
weight_dtype: 'fp16'  # [fp16, fp32]
enable_zero_snr: True
# pretrained model
base_model_path: '/content/FollowYourEmoji/ckpt_models/base/unet'
vae_model_path: '/content/FollowYourEmoji/ckpt_models/base/vae'
image_encoder_path: '/content/FollowYourEmoji/ckpt_models/base/image_encoder'
motion_module_path: '/content/FollowYourEmoji/ckpt_models/base/animatediff/mm_sd_v15_v2.ckpt'
# for infer
init_checkpoint: ''
init_num: 0
# model
model:
  unet_additional_kwargs:
    use_inflated_groupnorm: True
    unet_use_cross_frame_attention: False
    unet_use_temporal_attention: False
    use_motion_module: True
    motion_module_resolutions:
    - 1
    - 2
    - 4
    - 8
    motion_module_mid_block: True
    motion_module_decoder_only: False
    motion_module_type: Vanilla
    motion_module_kwargs:
      num_attention_heads: 8
      num_transformer_block: 1
      attention_block_types:
      - Temporal_Self
      - Temporal_Self
      temporal_position_encoding: True
      temporal_position_encoding_max_len: 32
      temporal_attention_dim_div: 1
    attention_mode: SpatialAtten # SpatialAtten/ReferOnlyAtten
  referencenet_additional_kwargs:
    info_mode: addRefImg
scheduler:
  num_train_timesteps: 1000
  beta_start:          0.00085
  beta_end:            0.012
  beta_schedule:       "scaled_linear"
  steps_offset:        1
  clip_sample:         false