merge

.github/workflows/pr_style_bot.yml

@@ -14,4 +14,4 @@ jobs:
     with:
       python_quality_dependencies: "[quality]"
     secrets:
-      bot_token: ${{ secrets.GITHUB_TOKEN }}
+      bot_token: ${{ secrets.HF_STYLE_BOT_ACTION }}

docs/source/en/api/models/chroma_transformer.md

@@ -0,0 +1,19 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# ChromaTransformer2DModel
+
+A modified flux Transformer model from [Chroma](https://huggingface.co/lodestones/Chroma)
+
+## ChromaTransformer2DModel
+
+[[autodoc]] ChromaTransformer2DModel

docs/source/en/api/pipelines/chroma.md

@@ -0,0 +1,70 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Chroma
+
+<div class="flex flex-wrap space-x-1">
+  <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+  <img alt="MPS" src="https://img.shields.io/badge/MPS-000000?style=flat&logo=apple&logoColor=white%22">
+</div>
+
+Chroma is a text to image generation model based on Flux.
+
+Original model checkpoints for Chroma can be found [here](https://huggingface.co/lodestones/Chroma).
+
+<Tip>
+
+Chroma can use all the same optimizations as Flux.
+
+
+## Inference (Single File)
+
+The `ChromaTransformer2DModel` supports loading checkpoints in the original format. This is also useful when trying to load finetunes or quantized versions of the models that have been published by the community.
+
+The following example demonstrates how to run Chroma from a single file.
+
+Then run the following example
+
+```python
+import torch
+from diffusers import ChromaTransformer2DModel, ChromaPipeline
+from transformers import T5EncoderModel
+
+bfl_repo = "black-forest-labs/FLUX.1-dev"
+dtype = torch.bfloat16
+
+transformer = ChromaTransformer2DModel.from_single_file("https://huggingface.co/lodestones/Chroma/blob/main/chroma-unlocked-v35.safetensors", torch_dtype=dtype)
+
+text_encoder = T5EncoderModel.from_pretrained(bfl_repo, subfolder="text_encoder_2", torch_dtype=dtype)
+tokenizer = T5Tokenizer.from_pretrained(bfl_repo, subfolder="tokenizer_2", torch_dtype=dtype)
+
+pipe = ChromaPipeline.from_pretrained(bfl_repo, transformer=transformer, text_encoder=text_encoder, tokenizer=tokenizer, torch_dtype=dtype)
+
+pipe.enable_model_cpu_offload()
+
+prompt = "A cat holding a sign that says hello world"
+image = pipe(
+    prompt,
+    guidance_scale=4.0,
+    output_type="pil",
+    num_inference_steps=26,
+    generator=torch.Generator("cpu").manual_seed(0)
+).images[0]
+
+image.save("image.png")
+```
+
+## ChromaPipeline
+
+[[autodoc]] ChromaPipeline
+	- all
+	- __call__

docs/source/en/quantization/bitsandbytes.md

@@ -416,6 +416,45 @@ text_encoder_2_4bit.dequantize()
 transformer_4bit.dequantize()
 ```
 
+## torch.compile
+
+Speed up inference with `torch.compile`. Make sure you have the latest `bitsandbytes` installed and we also recommend installing [PyTorch nightly](https://pytorch.org/get-started/locally/).
+
+<hfoptions id="bnb">
+<hfoption id="8-bit">
+```py
+torch._dynamo.config.capture_dynamic_output_shape_ops = True
+
+quant_config = DiffusersBitsAndBytesConfig(load_in_8bit=True)
+transformer_4bit = AutoModel.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    subfolder="transformer",
+    quantization_config=quant_config,
+    torch_dtype=torch.float16,
+)
+transformer_4bit.compile(fullgraph=True)
+```
+
+</hfoption>
+<hfoption id="4-bit">
+
+```py
+quant_config = DiffusersBitsAndBytesConfig(load_in_4bit=True)
+transformer_4bit = AutoModel.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    subfolder="transformer",
+    quantization_config=quant_config,
+    torch_dtype=torch.float16,
+)
+transformer_4bit.compile(fullgraph=True)
+```
+</hfoption>
+</hfoptions>
+
+On an RTX 4090 with compilation, 4-bit Flux generation completed in 25.809 seconds versus 32.570 seconds without.
+
+Check out the [benchmarking script](https://gist.github.com/sayakpaul/0db9d8eeeb3d2a0e5ed7cf0d9ca19b7d) for more details.
+
 ## Resources
 
 * [End-to-end notebook showing Flux.1 Dev inference in a free-tier Colab](https://gist.github.com/sayakpaul/c76bd845b48759e11687ac550b99d8b4)

docs/source/en/quantization/torchao.md

@@ -65,6 +65,9 @@ transformer = torch.compile(transformer, mode="max-autotune", fullgraph=True)
 
 For speed and memory benchmarks on Flux and CogVideoX, please refer to the table [here](https://github.com/huggingface/diffusers/pull/10009#issue-2688781450). You can also find some torchao [benchmarks](https://github.com/pytorch/ao/tree/main/torchao/quantization#benchmarks) numbers for various hardware.
 
+> [!TIP]
+> The FP8 post-training quantization schemes in torchao are effective for GPUs with compute capability of at least 8.9 (RTX-4090, Hopper, etc.). FP8 often provides the best speed, memory, and quality trade-off when generating images and videos. We recommend combining FP8 and torch.compile if your GPU is compatible.
+
 torchao also supports an automatic quantization API through [autoquant](https://github.com/pytorch/ao/blob/main/torchao/quantization/README.md#autoquantization). Autoquantization determines the best quantization strategy applicable to a model by comparing the performance of each technique on chosen input types and shapes. Currently, this can be used directly on the underlying modeling components. Diffusers will also expose an autoquant configuration option in the future.
 
 The `TorchAoConfig` class accepts three parameters:

src/diffusers/__init__.py

@@ -159,6 +159,7 @@
             "AutoencoderTiny",
             "AutoModel",
             "CacheMixin",
+            "ChromaTransformer2DModel",
             "CogVideoXTransformer3DModel",
             "CogView3PlusTransformer2DModel",
             "CogView4Transformer2DModel",
@@ -352,6 +353,7 @@
             "AuraFlowPipeline",
             "BlipDiffusionControlNetPipeline",
             "BlipDiffusionPipeline",
+            "ChromaPipeline",
             "CLIPImageProjection",
             "CogVideoXFunControlPipeline",
             "CogVideoXImageToVideoPipeline",
@@ -768,6 +770,7 @@
             AutoencoderTiny,
             AutoModel,
             CacheMixin,
+            ChromaTransformer2DModel,
             CogVideoXTransformer3DModel,
             CogView3PlusTransformer2DModel,
             CogView4Transformer2DModel,
@@ -940,6 +943,7 @@
             AudioLDM2UNet2DConditionModel,
             AudioLDMPipeline,
             AuraFlowPipeline,
+            ChromaPipeline,
             CLIPImageProjection,
             CogVideoXFunControlPipeline,
             CogVideoXImageToVideoPipeline,

src/diffusers/loaders/lora_pipeline.py

@@ -81,12 +81,17 @@ def _maybe_dequantize_weight_for_expanded_lora(model, module):
         from ..quantizers.gguf.utils import dequantize_gguf_tensor
 
     is_bnb_4bit_quantized = module.weight.__class__.__name__ == "Params4bit"
+    is_bnb_8bit_quantized = module.weight.__class__.__name__ == "Int8Params"
     is_gguf_quantized = module.weight.__class__.__name__ == "GGUFParameter"
 
     if is_bnb_4bit_quantized and not is_bitsandbytes_available():
         raise ValueError(
             "The checkpoint seems to have been quantized with `bitsandbytes` (4bits). Install `bitsandbytes` to load quantized checkpoints."
         )
+    if is_bnb_8bit_quantized and not is_bitsandbytes_available():
+        raise ValueError(
+            "The checkpoint seems to have been quantized with `bitsandbytes` (8bits). Install `bitsandbytes` to load quantized checkpoints."
+        )
     if is_gguf_quantized and not is_gguf_available():
         raise ValueError(
             "The checkpoint seems to have been quantized with `gguf`. Install `gguf` to load quantized checkpoints."
@@ -97,10 +102,10 @@ def _maybe_dequantize_weight_for_expanded_lora(model, module):
         weight_on_cpu = True
 
     device = torch.accelerator.current_accelerator().type if hasattr(torch, "accelerator") else "cuda"
-    if is_bnb_4bit_quantized:
+    if is_bnb_4bit_quantized or is_bnb_8bit_quantized:
         module_weight = dequantize_bnb_weight(
             module.weight.to(device) if weight_on_cpu else module.weight,
-            state=module.weight.quant_state,
+            state=module.weight.quant_state if is_bnb_4bit_quantized else module.state,
             dtype=model.dtype,
         ).data
     elif is_gguf_quantized:

src/diffusers/loaders/single_file_model.py

@@ -29,6 +29,7 @@
     convert_animatediff_checkpoint_to_diffusers,
     convert_auraflow_transformer_checkpoint_to_diffusers,
     convert_autoencoder_dc_checkpoint_to_diffusers,
+    convert_chroma_transformer_checkpoint_to_diffusers,
     convert_controlnet_checkpoint,
     convert_flux_transformer_checkpoint_to_diffusers,
     convert_hidream_transformer_to_diffusers,
@@ -97,6 +98,10 @@
         "checkpoint_mapping_fn": convert_flux_transformer_checkpoint_to_diffusers,
         "default_subfolder": "transformer",
     },
+    "ChromaTransformer2DModel": {
+        "checkpoint_mapping_fn": convert_chroma_transformer_checkpoint_to_diffusers,
+        "default_subfolder": "transformer",
+    },
     "LTXVideoTransformer3DModel": {
         "checkpoint_mapping_fn": convert_ltx_transformer_checkpoint_to_diffusers,
         "default_subfolder": "transformer",