Chroma pipeline documentation bug regarding the guidance_scale parameter #12359
Description
Describe the bug
From my understanding, Chroma is a retrained and dedistilled version of the Flux architecture, so it uses true CFG, unlike Flux. I can indeed confirm that this is true by tracing through the source code.
However, currently the documentation for the guidance_scale parameter in the ChromaPipeline.__call__() method mentions otherwise, presumably because it was copied over from the FluxPipeline documentation.
Reproduction
The current documentation for the guidance_scale parameter in the ChromaPipeline.__call__() method:
'''
guidance_scale (float, optional, defaults to 3.5) — Embedded guiddance scale is enabled by setting guidance_scale > 1. Higher guidance_scale encourages a model to generate images more aligned with prompt at the expense of lower image quality.
Guidance-distilled models approximates true classifer-free guidance for guidance_scale > 1. Refer to the [paper](https://huggingface.co/papers/2210.03142) to learn more.
'''Logs
System Info
- 🤗 Diffusers version: 0.36.0.dev0
- Platform: Windows-10-10.0.26100-SP0
- Running on Google Colab?: No
- Python version: 3.11.9
- PyTorch version (GPU?): 2.7.1+cu128 (True)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Huggingface_hub version: 0.34.4
- Transformers version: 4.55.0
- Accelerate version: 1.10.0
- PEFT version: 0.17.0
- Bitsandbytes version: 0.47.0
- Safetensors version: 0.6.2
- xFormers version: 0.0.31.post1
- Accelerator: NVIDIA GeForce RTX 4090, 24564 MiB
- Using GPU in script?: No
- Using distributed or parallel set-up in script?: No