luminars/config.py

"""Configuration dataclass for LuminaRS."""
from dataclasses import dataclass

@dataclass
class LuminaRSConfig:
    # Latent dims
    latent_dim: int          = 16      # VAE latent channels
    latent_h: int            = 32      # 32x32 latent = 1024x1024 px
    latent_w: int            = 32
    # UNet channels at each scale
    channels: tuple          = (64, 128, 256, 256, 384)
    # Bottleneck depth
    n_bottleneck: int        = 4
    # Iterative refinement depth (TRM-style shared-weight recursion)
    n_recurse: int           = 2
    # Time embedding
    t_embed_dim: int         = 256
    # Text conditioning
    text_embed_dim: int      = 768
    max_text_len: int        = 77
    # Training
    drop_path: float         = 0.05
    # VAE (frozen)
    vae_pretrained: str      = "madebyollin/sdxl-vae-fp16-fix"
    # Text encoder (frozen)
    clip_pretrained: str     = "openai/clip-vit-large-patch14"
    # Flow matching inference steps
    n_flow_steps: int        = 12