emb128/config.json

{
    "embed_dim": 256,
    "chip_vision_layers": [
        3,
        4,
        6,
        3
    ],
    "chip_vision_width": 64,
    "chip_vision_patch_size": null,
    "chip_transformer_width": 256,
    "chip_transformer_head": 8,
    "chip_transformer_layers": 6,
    "prior_depth": 6,
    "prior_dim_head": 64,
    "prior_heads": 8,
    "prior_sample_timesteps": 64,
    "prior_cond_drop_prob": 0.2,
    "decoder_unet_dim": 128,
    "decoder_unet_dim_mults": [
        1,
        2,
        4,
        8
    ],
    "decoder_image_cond_drop_prob": 0.1,
    "decoder_text_cond_drop_prob": 0.5,
    "decoder_latent_scaling_factor": 647.98193,
    "latent_n_channels": 8,
    "latent_image_resolution": 64,
    "decoder_auto_normalize_img": false,
    "decoder_learned_variance": false,
    "decoder_clip_denoised": false,
    "timesteps": 1000,
    "n_channels": 6,
    "image_resolution": 128,
    "n_features": 38
}