qwen-image-distill

README.md

@@ -362,10 +362,13 @@ https://github.com/Artiprocher/DiffSynth-Studio/assets/35051019/59fb2f7b-8de0-44
 
 
 ## Update History
-- **August 1, 2025** [FLUX.1-Krea-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Krea-dev) with a focus on aesthetic photography is comprehensively supported, including low-GPU-memory layer-by-layer offload, LoRA training and full training. See [./examples/flux/](./examples/flux/).
+
+- **August 5, 2025** We open-sourced the distilled acceleration model of Qwen-Image, [DiffSynth-Studio/Qwen-Image-Distill-Full](https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-Distill-Full), achieving approximately 5x speedup.
 
 - **August 4, 2025** 🔥 Qwen-Image is now open source. Welcome the new member to the image generation model family!
 
+- **August 1, 2025** [FLUX.1-Krea-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Krea-dev) with a focus on aesthetic photography is comprehensively supported, including low-GPU-memory layer-by-layer offload, LoRA training and full training. See [./examples/flux/](./examples/flux/).
+
 - **July 28, 2025** With the open-sourcing of Wan 2.2, we immediately provided comprehensive support, including low-GPU-memory layer-by-layer offload, FP8 quantization, sequence parallelism, LoRA training, full training. See [./examples/wanvideo/](./examples/wanvideo/).
 
 - **July 11, 2025** We propose Nexus-Gen, a unified model that synergizes the language reasoning capabilities of LLMs with the image synthesis power of diffusion models. This framework enables seamless image understanding, generation, and editing tasks.
@@ -375,13 +378,13 @@ https://github.com/Artiprocher/DiffSynth-Studio/assets/35051019/59fb2f7b-8de0-44
   - Training Dataset: [ModelScope Dataset](https://www.modelscope.cn/datasets/DiffSynth-Studio/Nexus-Gen-Training-Dataset)
   - Online Demo: [ModelScope Nexus-Gen Studio](https://www.modelscope.cn/studios/DiffSynth-Studio/Nexus-Gen)
 
+<details>
+<summary>More</summary>
+
 - **June 15, 2025** ModelScope's official evaluation framework, [EvalScope](https://github.com/modelscope/evalscope), now supports text-to-image generation evaluation. Try it with the [Best Practices](https://evalscope.readthedocs.io/zh-cn/latest/best_practice/t2i_eval.html) guide.
 
 - **March 25, 2025** Our new open-source project, [DiffSynth-Engine](https://github.com/modelscope/DiffSynth-Engine), is now open-sourced! Focused on stable model deployment. Geared towards industry. Offers better engineering support, higher computational performance, and more stable functionality.
 
-<details>
-<summary>More</summary>
-
 - **March 31, 2025** We support InfiniteYou, an identity preserving method for FLUX. Please refer to [./examples/InfiniteYou/](./examples/InfiniteYou/) for more details.
 
 - **March 13, 2025** We support HunyuanVideo-I2V, the image-to-video generation version of HunyuanVideo open-sourced by Tencent. Please refer to [./examples/HunyuanVideo/](./examples/HunyuanVideo/) for more details.

README_zh.md

@@ -92,6 +92,7 @@ image.save("image.jpg")
 |模型 ID|推理|全量训练|全量训练后验证|LoRA 训练|LoRA 训练后验证|
 |-|-|-|-|-|-|
 |[Qwen/Qwen-Image](https://www.modelscope.cn/models/Qwen/Qwen-Image)|[code](./examples/qwen_image/model_inference/Qwen-Image.py)|[code](./examples/qwen_image/model_training/full/Qwen-Image.sh)|[code](./examples/qwen_image/model_training/validate_full/Qwen-Image.py)|[code](./examples/qwen_image/model_training/lora/Qwen-Image.sh)|[code](./examples/qwen_image/model_training/validate_lora/Qwen-Image.py)|
+|[DiffSynth-Studio/Qwen-Image-Distill-Full](https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-Distill-Full)|[code](./examples/qwen_image/model_inference/Qwen-Image-Distill-Full.py)|[code](./examples/qwen_image/model_training/full/Qwen-Image-Distill-Full.sh)|[code](./examples/qwen_image/model_training/validate_full/Qwen-Image-Distill-Full.py)|[code](./examples/qwen_image/model_training/lora/Qwen-Image-Distill-Full.sh)|[code](./examples/qwen_image/model_training/validate_lora/Qwen-Image-Distill-Full.py)|
 
 </details>
 
@@ -378,10 +379,13 @@ https://github.com/Artiprocher/DiffSynth-Studio/assets/35051019/59fb2f7b-8de0-44
 
 
 ## 更新历史
-- **2025年8月1日** [FLUX.1-Krea-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Krea-dev) 开源，这是一个专注于美学摄影的文生图模型。我们第一时间提供了全方位支持，包括低显存逐层 offload、LoRA 训练、全量训练。详细信息请参考 [./examples/flux/](./examples/flux/)。
+
+- **2025年8月5日** 我们开源了 Qwen-Image 的蒸馏加速模型 [DiffSynth-Studio/Qwen-Image-Distill-Full](https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-Distill-Full)，实现了约 5 倍加速。
 
 - **2025年8月4日** 🔥 Qwen-Image 开源，欢迎图像生成模型家族新成员！
 
+- **2025年8月1日** [FLUX.1-Krea-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Krea-dev) 开源，这是一个专注于美学摄影的文生图模型。我们第一时间提供了全方位支持，包括低显存逐层 offload、LoRA 训练、全量训练。详细信息请参考 [./examples/flux/](./examples/flux/)。
+
 - **2025年7月28日** Wan 2.2 开源，我们第一时间提供了全方位支持，包括低显存逐层 offload、FP8 量化、序列并行、LoRA 训练、全量训练。详细信息请参考 [./examples/wanvideo/](./examples/wanvideo/)。
 
 - **2025年7月11日** 我们提出 Nexus-Gen，一个将大语言模型（LLM）的语言推理能力与扩散模型的图像生成能力相结合的统一框架。该框架支持无缝的图像理解、生成和编辑任务。
@@ -391,13 +395,13 @@ https://github.com/Artiprocher/DiffSynth-Studio/assets/35051019/59fb2f7b-8de0-44
   - 训练数据集: [ModelScope Dataset](https://www.modelscope.cn/datasets/DiffSynth-Studio/Nexus-Gen-Training-Dataset)
   - 在线体验: [ModelScope Nexus-Gen Studio](https://www.modelscope.cn/studios/DiffSynth-Studio/Nexus-Gen)
 
+<details>
+<summary>更多</summary>
+
 - **2025年6月15日** ModelScope 官方评测框架 [EvalScope](https://github.com/modelscope/evalscope) 现已支持文生图生成评测。请参考[最佳实践](https://evalscope.readthedocs.io/zh-cn/latest/best_practice/t2i_eval.html)指南进行尝试。
 
 - **2025年3月25日** 我们的新开源项目 [DiffSynth-Engine](https://github.com/modelscope/DiffSynth-Engine) 现已开源！专注于稳定的模型部署，面向工业界，提供更好的工程支持、更高的计算性能和更稳定的功能。
 
-<details>
-<summary>更多</summary>
-
 - **2025年3月31日** 我们支持 InfiniteYou，一种用于 FLUX 的人脸特征保留方法。更多细节请参考 [./examples/InfiniteYou/](./examples/InfiniteYou/)。
 
 - **2025年3月13日** 我们支持 HunyuanVideo-I2V，即腾讯开源的 HunyuanVideo 的图像到视频生成版本。更多细节请参考 [./examples/HunyuanVideo/](./examples/HunyuanVideo/)。

examples/qwen_image/README.md

@@ -43,6 +43,7 @@ image.save("image.jpg")
 |Model ID|Inference|Full Training|Validation after Full Training|LoRA Training|Validation after LoRA Training|
 |-|-|-|-|-|-|
 |[Qwen/Qwen-Image](https://www.modelscope.cn/models/Qwen/Qwen-Image  )|[code](./model_inference/Qwen-Image.py)|[code](./model_training/full/Qwen-Image.sh)|[code](./model_training/validate_full/Qwen-Image.py)|[code](./model_training/lora/Qwen-Image.sh)|[code](./model_training/validate_lora/Qwen-Image.py)|
+|[DiffSynth-Studio/Qwen-Image-Distill-Full](https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-Distill-Full)|[code](./model_inference/Qwen-Image-Distill-Full.py)|[code](./model_training/full/Qwen-Image-Distill-Full.sh)|[code](./model_training/validate_full/Qwen-Image-Distill-Full.py)|[code](./model_training/lora/Qwen-Image-Distill-Full.sh)|[code](./model_training/validate_lora/Qwen-Image-Distill-Full.py)|
 
 
 ## Model Inference

examples/qwen_image/README_zh.md

@@ -43,6 +43,7 @@ image.save("image.jpg")
 |模型 ID|推理|全量训练|全量训练后验证|LoRA 训练|LoRA 训练后验证|
 |-|-|-|-|-|-|
 |[Qwen/Qwen-Image](https://www.modelscope.cn/models/Qwen/Qwen-Image)|[code](./model_inference/Qwen-Image.py)|[code](./model_training/full/Qwen-Image.sh)|[code](./model_training/validate_full/Qwen-Image.py)|[code](./model_training/lora/Qwen-Image.sh)|[code](./model_training/validate_lora/Qwen-Image.py)|
+|[DiffSynth-Studio/Qwen-Image-Distill-Full](https://www.modelscope.cn/models/DiffSynth-Studio/Qwen-Image-Distill-Full)|[code](./model_inference/Qwen-Image-Distill-Full.py)|[code](./model_training/full/Qwen-Image-Distill-Full.sh)|[code](./model_training/validate_full/Qwen-Image-Distill-Full.py)|[code](./model_training/lora/Qwen-Image-Distill-Full.sh)|[code](./model_training/validate_lora/Qwen-Image-Distill-Full.py)|
 
 
 ## 模型推理

examples/qwen_image/model_inference/Qwen-Image-Distill-Full.py

@@ -0,0 +1,17 @@
+from diffsynth.pipelines.qwen_image import QwenImagePipeline, ModelConfig
+import torch
+
+
+pipe = QwenImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="DiffSynth-Studio/Qwen-Image-Distill-Full", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors"),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
+    ],
+    tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
+)
+prompt = "精致肖像，水下少女，蓝裙飘逸，发丝轻扬，光影透澈，气泡环绕，面容恬静，细节精致，梦幻唯美。"
+image = pipe(prompt, seed=0, num_inference_steps=15, cfg_scale=1)
+image.save("image.jpg")

examples/qwen_image/model_inference_lor_vram/Qwen-Image-Distill-Full.py

@@ -0,0 +1,18 @@
+from diffsynth.pipelines.qwen_image import QwenImagePipeline, ModelConfig
+import torch
+
+
+pipe = QwenImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="DiffSynth-Studio/Qwen-Image-Distill-Full", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+    ],
+    tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
+)
+pipe.enable_vram_management()
+prompt = "精致肖像，水下少女，蓝裙飘逸，发丝轻扬，光影透澈，气泡环绕，面容恬静，细节精致，梦幻唯美。"
+image = pipe(prompt, seed=0, num_inference_steps=15, cfg_scale=1)
+image.save("image.jpg")

examples/qwen_image/model_training/full/Qwen-Image-Distill-Full.sh

@@ -0,0 +1,12 @@
+accelerate launch --config_file examples/qwen_image/model_training/full/accelerate_config_zero2offload.yaml examples/qwen_image/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata.csv \
+  --max_pixels 1048576 \
+  --dataset_repeat 50 \
+  --model_id_with_origin_paths "DiffSynth-Studio/Qwen-Image-Distill-Full:diffusion_pytorch_model*.safetensors,Qwen/Qwen-Image:text_encoder/model*.safetensors,Qwen/Qwen-Image:vae/diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-5 \
+  --num_epochs 2 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/Qwen-Image-Distill-Full_full" \
+  --trainable_models "dit" \
+  --use_gradient_checkpointing

examples/qwen_image/model_training/lora/Qwen-Image-Distill-Full.sh

@@ -0,0 +1,15 @@
+accelerate launch examples/qwen_image/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata.csv \
+  --max_pixels 1048576 \
+  --dataset_repeat 50 \
+  --model_id_with_origin_paths "DiffSynth-Studio/Qwen-Image-Distill-Full:diffusion_pytorch_model*.safetensors,Qwen/Qwen-Image:text_encoder/model*.safetensors,Qwen/Qwen-Image:vae/diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/Qwen-Image-Distill-Full_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "to_q,to_k,to_v,add_q_proj,add_k_proj,add_v_proj,to_out.0,to_add_out,img_mlp.net.2,img_mod.1,txt_mlp.net.2,txt_mod.1" \
+  --lora_rank 32 \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/qwen_image/model_training/validate_full/Qwen-Image-Distill-Full.py

@@ -0,0 +1,20 @@
+from diffsynth.pipelines.qwen_image import QwenImagePipeline, ModelConfig
+from diffsynth import load_state_dict
+import torch
+
+
+pipe = QwenImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="DiffSynth-Studio/Qwen-Image-Distill-Full", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors"),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
+    ],
+    tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
+)
+state_dict = load_state_dict("models/train/Qwen-Image-Distill-Full_full/epoch-1.safetensors")
+pipe.dit.load_state_dict(state_dict)
+prompt = "a dog"
+image = pipe(prompt, seed=0, num_inference_steps=15, cfg_scale=1)
+image.save("image.jpg")

examples/qwen_image/model_training/validate_full/Qwen-Image.py

@@ -7,9 +7,9 @@ pipe = QwenImagePipeline.from_pretrained(
     torch_dtype=torch.bfloat16,
     device="cuda",
     model_configs=[
-        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors"),
-        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors"),
-        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
     ],
     tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
 )

examples/qwen_image/model_training/validate_lora/Qwen-Image-Distill-Full.py

@@ -0,0 +1,18 @@
+from diffsynth.pipelines.qwen_image import QwenImagePipeline, ModelConfig
+import torch
+
+
+pipe = QwenImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="DiffSynth-Studio/Qwen-Image-Distill-Full", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors"),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
+    ],
+    tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
+)
+pipe.load_lora(pipe.dit, "models/train/Qwen-Image-Distill-Full_lora/epoch-4.safetensors")
+prompt = "a dog"
+image = pipe(prompt, seed=0, num_inference_steps=15, cfg_scale=1)
+image.save("image.jpg")

examples/qwen_image/model_training/validate_lora/Qwen-Image.py

@@ -6,9 +6,9 @@ pipe = QwenImagePipeline.from_pretrained(
     torch_dtype=torch.bfloat16,
     device="cuda",
     model_configs=[
-        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors"),
-        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="text_encoder/model*.safetensors"),
-        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors", offload_device="cpu", offload_dtype=torch.float8_e4m3fn),
+        ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
     ],
     tokenizer_config=ModelConfig(model_id="Qwen/Qwen-Image", origin_file_pattern="tokenizer/"),
 )