| --- |
| base_model: black-forest-labs/FLUX.1-dev |
| library_name: diffusers |
| license: other |
| license_name: flux-1-dev-non-commercial-license |
| license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md |
| pipeline_tag: text-to-image |
| tags: |
| - flux |
| - text-to-image |
| - quantized |
| - fp8 |
| - e4m3 |
| - diffusers |
| base_model_relation: quantized |
| --- |
| # FLUX.1-dev-e4m3 |
|
|
| FP8 (e4m3) dynamically-quantized [FLUX.1-dev](https://huggingface.co/black-forest-labs/FLUX.1-dev), |
| saved as a complete `FluxPipeline`. |
|
|
| ## What was changed |
|
|
| Every double and single transformer block of the `FluxTransformer2DModel` is |
| quantized to `e4m3_e4m3_dynamic` — `float8_e4m3fn` weights with dynamically |
| scaled `float8_e4m3fn` activations. The rest of the pipeline is unchanged: the |
| transformer's non-block tensors, the CLIP text encoder and the VAE stay in |
| fp16, and the T5 text encoder stays in bf16. The transformer shrinks from |
| ~23.8 GB to ~12.0 GB. |
|
|
| This is the same recipe as |
| [`wavespeed/FLUX.1-dev-int8`](https://huggingface.co/wavespeed/FLUX.1-dev-int8) |
| with an fp8 rather than int8 numeric format. FP8 matmul needs Hopper (H100/H200) |
| or newer; on Ada and older the weights dequantize instead and you lose the speedup. |
|
|
| Quantization was done with WaveSpeed's `xelerate.ao.quantize`. Weights are |
| stored as pickled `.bin` shards, so loading requires `use_safetensors=False`. |
|
|
| ## Usage |
|
|
| ```python |
| import torch |
| from diffusers import FluxPipeline |
| |
| pipe = FluxPipeline.from_pretrained( |
| "wavespeed/FLUX.1-dev-e4m3", |
| torch_dtype=torch.float16, |
| use_safetensors=False, |
| ).to("cuda") |
| ``` |
|
|
| ## License |
|
|
| Derived from FLUX.1-dev, so the |
| [FLUX.1 \[dev\] Non-Commercial License](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md) |
| applies to these weights and to anything generated with them. Not for |
| commercial use. |
|
|
