VOSR_1.4B_os/args.json

{
    "resolution": 512,
    "patch_size": 2,
    "mlp_ratio": 4,
    "use_qknorm": true,
    "use_swiglu": true,
    "use_rope": true,
    "use_rmsnorm": true,
    "wo_shift": false,
    "dim": 1536,
    "depth": 36,
    "num_heads": 24,
    "ae_type": "qwen",
    "ae_path": "preset/ckpts/Qwen-Image-vae-2d",
    "time_dist": ["uniform", 0, 1.0],
    "cfg_ratio": 0.1,
    "cfg_scale": 0.5,
    "dinov2_size": 448,
    "enc_type": "dinov2l",
    "enc_dim": 1024,
    "layer_dinov2b_list": [17],
    "interp_type": "lin",
    "encdim_ratio": 3,
    "weak_cond_strength_aelq_list": [0.05, 0.25],
    "weak_cond_strength_aelq": 0.1,
    "cond_strength_aelq_list": [5, 1],
    "t_start": 0,
    "t_end": 1,
    "distill_type": "rcgm"
}