mingyi456/Z-Image-Turbo-Misc-Finetunes-DF11

These are DF11 conversions of various Z-Image Turbo finetunes which I cannot find the corresponding source repo on Hugging Face.

Filename	Link to source model
redcraftRedzimageUpdatedDEC03_redzimage15AIO-DF11.safetensors	https://civitai.com/models/958009?modelVersionId=2462789
darkBeastZDBZUpdatedDEC21_dbzAIOV10-DF11.safetensors	https://civitai.com/models/2242173?modelVersionId=2524092
zimageTurboNSFWBy_labRatV0-DF11.safetensors	https://civitai.com/models/2221503?modelVersionId=2500972
zMania_alpha-DF11.safetensors	https://civitai.com/models/2231360?modelVersionId=2511906
zMania_beta-DF11.safetensors	https://civitai.com/models/2231360?modelVersionId=2543332
jibMixZIT_v10-DF11.safetensors	https://civitai.com/models/2231351?modelVersionId=2511897
jibMixZIT_v20-DF11.safetensors	https://civitai.com/models/2231351?modelVersionId=2637947
darkBeastFeb1226Latest_darkblitzZIT5steps-DF11.safetensors	https://civitai.com/models/2242173?modelVersionId=2682031
intorealism_zitV10-DF11.safetensors	https://civitai.com/models/1609320?modelVersionId=2740834
intorealism_zitV20-DF11.safetensors	https://civitai.com/models/1609320?modelVersionId=2790469
harukiMIX_zit2602-DF11.safetensors	https://civitai.com/models/856375?modelVersionId=2675194
harukiMIX_zit2603-DF11.safetensors	https://civitai.com/models/856375?modelVersionId=2815582
divingZImageTurbo_v30Fp16-DF11.safetensors	https://civitai.com/models/2276359?modelVersionId=2602310
divingZImageTurbo_v40Fp16-DF11.safetensors	https://civitai.com/models/2276359?modelVersionId=2633987
divingZImageTurbo_v50Fp16-DF11.safetensors	https://civitai.com/models/2276359?modelVersionId=2661770
gonzalomoZpop_v30AIO-DF11.safetensors	https://civitai.com/models/2192562?modelVersionId=2598924
gonzalomoZpop_insta-DF11.safetensors	https://civitai.com/models/2192562?modelVersionId=2780807
zImageTurbo_v10-DF11.safetensors	https://civitai.red/models/2538147/z-image-turbo-deedeemegadoodo-edition?modelVersionId=2852497&sync-account=green

For more information (including how to compress models yourself), check out https://huggingface.co/DFloat11 and https://github.com/LeanModels/DFloat11

Feel free to request for other models for compression as well (for either the diffusers library, ComfyUI, or any other model), although models that use architectures which are unfamiliar to me might be more difficult.

How to Use

diffusers

import torch
from diffusers import ZImagePipeline, ZImageTransformer2DModel
from dfloat11 import DFloat11Model
# from transformers.modeling_utils import no_init_weights # for transformers<5.0.0
from transformers.initialization import no_init_weights # for transformers>=5.0.0
pattern_dict = {
    r"noise_refiner\.\d+": (
        "attention.to_q",
        "attention.to_k",
        "attention.to_v",
        "attention.to_out.0",
        "feed_forward.w1",
        "feed_forward.w2",
        "feed_forward.w3",
        "adaLN_modulation.0"
    ),
    r"context_refiner\.\d+": (
        "attention.to_q",
        "attention.to_k",
        "attention.to_v",
        "attention.to_out.0",
        "feed_forward.w1",
        "feed_forward.w2",
        "feed_forward.w3",
    ),
    r"layers\.\d+": (
        "attention.to_q",
        "attention.to_k",
        "attention.to_v",
        "attention.to_out.0",
        "feed_forward.w1",
        "feed_forward.w2",
        "feed_forward.w3",
        "adaLN_modulation.0"
    ),
    r"cap_embedder": (
        "1",
    )
}
text_encoder = DFloat11Model.from_pretrained("DFloat11/Qwen3-4B-DF11", device="cpu")
with no_init_weights():
    transformer = ZImageTransformer2DModel.from_config(
        ZImageTransformer2DModel.load_config(
            "Tongyi-MAI/Z-Image-Turbo", subfolder="transformer"
        ),
        torch_dtype=torch.bfloat16
    ).to(torch.bfloat16)
# Make sure to download the file first, and edit the filepath accordingly
DFloat11Model.from_single_file(
    r".\zMania_beta-DF11.safetensors",
    device='cpu', 
    bfloat16_model=transformer, 
    pattern_dict=pattern_dict
)
pipe = ZImagePipeline.from_pretrained(
    "Tongyi-MAI/Z-Image-Turbo",
    text_encoder=text_encoder,
    transformer=transformer,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=False,
)
pipe.to("cuda")

ComfyUI

Refer to this model instead.

Compression details

This is the pattern_dict for compression:

pattern_dict = {
    r"noise_refiner\.\d+": (
        "attention.to_q",
        "attention.to_k",
        "attention.to_v",
        "attention.to_out.0",
        "feed_forward.w1",
        "feed_forward.w2",
        "feed_forward.w3",
        "adaLN_modulation.0"
    ),
    r"context_refiner\.\d+": (
        "attention.to_q",
        "attention.to_k",
        "attention.to_v",
        "attention.to_out.0",
        "feed_forward.w1",
        "feed_forward.w2",
        "feed_forward.w3",
    ),
    r"layers\.\d+": (
        "attention.to_q",
        "attention.to_k",
        "attention.to_v",
        "attention.to_out.0",
        "feed_forward.w1",
        "feed_forward.w2",
        "feed_forward.w3",
        "adaLN_modulation.0"
    ),
    r"cap_embedder": (
        "1",
    )
}