---
license: apache-2.0
language:
- en
pipeline_tag: text-to-image
tags:
- diffusion-single-file
base_model:
- black-forest-labs/FLUX.2-klein-4B
base_model_relation: quantized
library_name: diffusers
---
## Note: It appears that the bug with ComfyUI and DF11 FLUX.2 [klein] models does not occur in `diffusers`, I am still investigating the issue with the ComfyUI implementation.

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`

```python
import torch
from diffusers import Flux2KleinPipeline, Flux2Transformer2DModel
from transformers.modeling_utils import no_init_weights
from dfloat11 import DFloat11Model

text_encoder = DFloat11Model.from_pretrained("DFloat11/Qwen3-4B-DF11", device="cpu")

with no_init_weights():
	transformer = Flux2Transformer2DModel.from_config(
		Flux2Transformer2DModel.load_config(
			"black-forest-labs/FLUX.2-klein-4B", subfolder="transformer"
		),
		torch_dtype=torch.bfloat16
	).to(torch.bfloat16)
DFloat11Model.from_pretrained("mingyi456/FLUX.2-klein-4B-DF11", device="cpu", bfloat16_model=transformer)

pipe = Flux2KleinPipeline.from_pretrained("black-forest-labs/FLUX.2-klein-4B", text_encoder=text_encoder, transformer=transformer, torch_dtype=torch.bfloat16)
pipe.to("cuda")

prompt = "A cat holding a sign that says hello world"
image = pipe(
    prompt=prompt,
    height=1024,
    width=1024,
    guidance_scale=1.0,
    num_inference_steps=4,
    generator=torch.Generator(device=device).manual_seed(0)
).images[0]
image.save("flux-klein.png")
```

#### ComfyUI
Refer to this [model](https://huggingface.co/mingyi456/FLUX.2-klein-4B-DF11-ComfyUI) instead.

### Compression details

This is the `pattern_dict` for compression:

```python
pattern_dict = {
    r"double_stream_modulation_img.linear": [],
    r"double_stream_modulation_txt.linear": [],
    r"single_stream_modulation.linear": [],
    r"context_embedder": [],
    r"transformer_blocks\.\d+" : (
        "attn.to_q",
        "attn.to_k",
        "attn.to_v",
        "attn.to_out.0",
        "attn.add_q_proj",
        "attn.add_k_proj",
        "attn.add_v_proj",
        "attn.to_add_out",
        "ff.linear_in",
        "ff.linear_out",
        "ff_context.linear_in",
        "ff_context.linear_out",
    ),
    r"single_transformer_blocks\.\d+" : (
        "attn.to_qkv_mlp_proj",
        "attn.to_out",
    ),
    r"norm_out.linear": [],
}
```