support flux any training

examples/flux/model_training/full/FLEX.2-preview.sh

@@ -0,0 +1,12 @@
+accelerate launch --config_file examples/flux/model_training/full/accelerate_config.yaml examples/flux/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 200 \
+  --model_id_with_origin_paths "ostris/Flex.2-preview:Flex.2-preview.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors" \
+  --learning_rate 1e-5 \
+  --num_epochs 1 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/FLEX.2-preview_full" \
+  --trainable_models "dit" \
+  --use_gradient_checkpointing

examples/flux/model_training/full/FLUX.1-dev-Controlnet-Inpainting-Beta.sh

@@ -0,0 +1,14 @@
+accelerate launch --config_file examples/flux/model_training/full/accelerate_config.yaml examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_controlnet_inpaint.csv \
+  --data_file_keys "image,controlnet_image,controlnet_inpaint_mask" \
+  --max_pixels 1048576 \
+  --dataset_repeat 400 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta:diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-5 \
+  --num_epochs 1 \
+  --remove_prefix_in_ckpt "pipe.controlnet.models.0." \
+  --output_path "./models/train/FLUX.1-dev-Controlnet-Inpainting-Beta_full" \
+  --trainable_models "controlnet" \
+  --extra_inputs "controlnet_image,controlnet_inpaint_mask" \
+  --use_gradient_checkpointing

examples/flux/model_training/full/FLUX.1-dev-Controlnet-Union-alpha.sh

@@ -0,0 +1,14 @@
+accelerate launch --config_file examples/flux/model_training/full/accelerate_config.yaml examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_controlnet_canny.csv \
+  --data_file_keys "image,controlnet_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 400 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,InstantX/FLUX.1-dev-Controlnet-Union-alpha:diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-5 \
+  --num_epochs 1 \
+  --remove_prefix_in_ckpt "pipe.controlnet.models.0." \
+  --output_path "./models/train/FLUX.1-dev-Controlnet-Union-alpha_full" \
+  --trainable_models "controlnet" \
+  --extra_inputs "controlnet_image,controlnet_processor_id" \
+  --use_gradient_checkpointing

examples/flux/model_training/full/FLUX.1-dev-Controlnet-Upscaler.sh

@@ -0,0 +1,14 @@
+accelerate launch --config_file examples/flux/model_training/full/accelerate_config.yaml examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_controlnet_upscale.csv \
+  --data_file_keys "image,controlnet_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 400 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,jasperai/Flux.1-dev-Controlnet-Upscaler:diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-5 \
+  --num_epochs 1 \
+  --remove_prefix_in_ckpt "pipe.controlnet.models.0." \
+  --output_path "./models/train/FLUX.1-dev-Controlnet-Upscaler_full" \
+  --trainable_models "controlnet" \
+  --extra_inputs "controlnet_image" \
+  --use_gradient_checkpointing

examples/flux/model_training/full/FLUX.1-dev-InfiniteYou.sh

@@ -0,0 +1,14 @@
+accelerate launch --config_file examples/flux/model_training/full/accelerate_config.yaml examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_infiniteyou.csv \
+  --data_file_keys "image,controlnet_image,infinityou_id_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 400 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,ByteDance/InfiniteYou:infu_flux_v1.0/aes_stage2/image_proj_model.bin,ByteDance/InfiniteYou:infu_flux_v1.0/aes_stage2/InfuseNetModel/*.safetensors" \
+  --learning_rate 1e-5 \
+  --num_epochs 1 \
+  --remove_prefix_in_ckpt "pipe." \
+  --output_path "./models/train/FLUX.1-dev-InfiniteYou_full" \
+  --trainable_models "controlnet,image_proj_model" \
+  --extra_inputs "controlnet_image,infinityou_id_image,infinityou_guidance" \
+  --use_gradient_checkpointing

examples/flux/model_training/full/Step1X-Edit.sh

@@ -0,0 +1,14 @@
+accelerate launch --config_file examples/flux/model_training/full/accelerate_config.yaml examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_step1x.csv \
+  --data_file_keys "image,step1x_reference_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 400 \
+  --model_id_with_origin_paths "Qwen/Qwen2.5-VL-7B-Instruct:,stepfun-ai/Step1X-Edit:step1x-edit-i1258.safetensors,stepfun-ai/Step1X-Edit:vae.safetensors" \
+  --learning_rate 1e-5 \
+  --num_epochs 1 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/Step1X-Edit_full" \
+  --trainable_models "dit" \
+  --extra_inputs "step1x_reference_image" \
+  --use_gradient_checkpointing_offload

examples/flux/model_training/lora/FLEX.2-preview.sh

@@ -0,0 +1,15 @@
+accelerate launch examples/flux/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 "ostris/Flex.2-preview:Flex.2-preview.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/FLEX.2-preview_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "a_to_qkv,b_to_qkv,ff_a.0,ff_a.2,ff_b.0,ff_b.2,a_to_out,b_to_out,proj_out,norm.linear,norm1_a.linear,norm1_b.linear,to_qkv_mlp" \
+  --lora_rank 32 \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Inpainting-Beta.sh

@@ -0,0 +1,17 @@
+accelerate launch examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_controlnet_inpaint.csv \
+  --data_file_keys "image,controlnet_image,controlnet_inpaint_mask" \
+  --max_pixels 1048576 \
+  --dataset_repeat 100 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta:diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/FLUX.1-dev-Controlnet-Inpainting-Beta_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "a_to_qkv,b_to_qkv,ff_a.0,ff_a.2,ff_b.0,ff_b.2,a_to_out,b_to_out,proj_out,norm.linear,norm1_a.linear,norm1_b.linear,to_qkv_mlp" \
+  --lora_rank 32 \
+  --extra_inputs "controlnet_image,controlnet_inpaint_mask" \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Union-alpha.sh

@@ -0,0 +1,17 @@
+accelerate launch examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_controlnet_canny.csv \
+  --data_file_keys "image,controlnet_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 100 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,InstantX/FLUX.1-dev-Controlnet-Union-alpha:diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/FLUX.1-dev-Controlnet-Union-alpha_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "a_to_qkv,b_to_qkv,ff_a.0,ff_a.2,ff_b.0,ff_b.2,a_to_out,b_to_out,proj_out,norm.linear,norm1_a.linear,norm1_b.linear,to_qkv_mlp" \
+  --lora_rank 32 \
+  --extra_inputs "controlnet_image,controlnet_processor_id" \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Upscaler.sh

@@ -0,0 +1,17 @@
+accelerate launch examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_controlnet_upscale.csv \
+  --data_file_keys "image,controlnet_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 100 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,jasperai/Flux.1-dev-Controlnet-Upscaler:diffusion_pytorch_model.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/FLUX.1-dev-Controlnet-Upscaler_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "a_to_qkv,b_to_qkv,ff_a.0,ff_a.2,ff_b.0,ff_b.2,a_to_out,b_to_out,proj_out,norm.linear,norm1_a.linear,norm1_b.linear,to_qkv_mlp" \
+  --lora_rank 32 \
+  --extra_inputs "controlnet_image" \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/flux/model_training/lora/FLUX.1-dev-IP-Adapter.sh

@@ -0,0 +1,17 @@
+accelerate launch examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_ipadapter.csv \
+  --data_file_keys "image,ipadapter_images" \
+  --max_pixels 1048576 \
+  --dataset_repeat 50 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,InstantX/FLUX.1-dev-IP-Adapter:ip-adapter.bin,google/siglip-so400m-patch14-384:" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/FLUX.1-dev-IP-Adapter_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "a_to_qkv,b_to_qkv,ff_a.0,ff_a.2,ff_b.0,ff_b.2,a_to_out,b_to_out,proj_out,norm.linear,norm1_a.linear,norm1_b.linear,to_qkv_mlp" \
+  --lora_rank 32 \
+  --extra_inputs "ipadapter_images" \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/flux/model_training/lora/FLUX.1-dev-InfiniteYou.sh

@@ -0,0 +1,17 @@
+accelerate launch examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_infiniteyou.csv \
+  --data_file_keys "image,controlnet_image,infinityou_id_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 100 \
+  --model_id_with_origin_paths "black-forest-labs/FLUX.1-dev:flux1-dev.safetensors,black-forest-labs/FLUX.1-dev:text_encoder/model.safetensors,black-forest-labs/FLUX.1-dev:text_encoder_2/,black-forest-labs/FLUX.1-dev:ae.safetensors,ByteDance/InfiniteYou:infu_flux_v1.0/aes_stage2/image_proj_model.bin,ByteDance/InfiniteYou:infu_flux_v1.0/aes_stage2/InfuseNetModel/*.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/FLUX.1-dev-InfiniteYou_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "a_to_qkv,b_to_qkv,ff_a.0,ff_a.2,ff_b.0,ff_b.2,a_to_out,b_to_out,proj_out,norm.linear,norm1_a.linear,norm1_b.linear,to_qkv_mlp" \
+  --lora_rank 32 \
+  --extra_inputs "controlnet_image,infinityou_id_image,infinityou_guidance" \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/flux/model_training/lora/Step1X-Edit.sh

@@ -0,0 +1,17 @@
+accelerate launch examples/flux/model_training/train.py \
+  --dataset_base_path data/example_image_dataset \
+  --dataset_metadata_path data/example_image_dataset/metadata_step1x.csv \
+  --data_file_keys "image,step1x_reference_image" \
+  --max_pixels 1048576 \
+  --dataset_repeat 50 \
+  --model_id_with_origin_paths "Qwen/Qwen2.5-VL-7B-Instruct:,stepfun-ai/Step1X-Edit:step1x-edit-i1258.safetensors,stepfun-ai/Step1X-Edit:vae.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/Step1X-Edit_lora" \
+  --lora_base_model "dit" \
+  --lora_target_modules "a_to_qkv,b_to_qkv,ff_a.0,ff_a.2,ff_b.0,ff_b.2,a_to_out,b_to_out,proj_out,norm.linear,norm1_a.linear,norm1_b.linear,to_qkv_mlp" \
+  --lora_rank 32 \
+  --extra_inputs "step1x_reference_image" \
+  --align_to_opensource_format \
+  --use_gradient_checkpointing

examples/flux/model_training/train.py

@@ -1,5 +1,5 @@
 import torch, os, json
-from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
 from diffsynth.trainers.utils import DiffusionTrainingModule, ImageDataset, ModelLogger, launch_training_task, flux_parser
 from diffsynth.models.lora import FluxLoRAConverter
 os.environ["TOKENIZERS_PARALLELISM"] = "false"
@@ -51,7 +51,7 @@ class FluxTrainingModule(DiffusionTrainingModule):
     def forward_preprocess(self, data):
         # CFG-sensitive parameters
         inputs_posi = {"prompt": data["prompt"]}
-        inputs_nega = {}
+        inputs_nega = {"negative_prompt": ""}
         
         # CFG-unsensitive parameters
         inputs_shared = {
@@ -72,8 +72,14 @@ class FluxTrainingModule(DiffusionTrainingModule):
         }
         
         # Extra inputs
+        controlnet_input = {}
         for extra_input in self.extra_inputs:
-            inputs_shared[extra_input] = data[extra_input]
+            if extra_input.startswith("controlnet_"):
+                controlnet_input[extra_input.replace("controlnet_", "")] = data[extra_input]
+            else:
+                inputs_shared[extra_input] = data[extra_input]
+        if len(controlnet_input) > 0:
+            inputs_shared["controlnet_inputs"] = [ControlNetInput(**controlnet_input)]
         
         # Pipeline units will automatically process the input parameters.
         for unit in self.pipe.units:
@@ -100,6 +106,7 @@ if __name__ == "__main__":
         lora_base_model=args.lora_base_model,
         lora_target_modules=args.lora_target_modules,
         lora_rank=args.lora_rank,
+        use_gradient_checkpointing=args.use_gradient_checkpointing,
         use_gradient_checkpointing_offload=args.use_gradient_checkpointing_offload,
         extra_inputs=args.extra_inputs,
     )

examples/flux/model_training/validate_full/FLEX.2-preview.py

@@ -0,0 +1,20 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig
+from diffsynth import load_state_dict
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="ostris/Flex.2-preview", origin_file_pattern="Flex.2-preview.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+    ],
+)
+state_dict = load_state_dict("models/train/FLEX.2-preview_full/epoch-0.safetensors")
+pipe.dit.load_state_dict(state_dict)
+
+image = pipe(prompt="dog,white and brown dog, sitting on wall, under pink flowers", seed=0)
+image.save("image_FLEX.2-preview_full.jpg")

examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Inpainting-Beta.py

@@ -0,0 +1,31 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from diffsynth import load_state_dict
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta", origin_file_pattern="diffusion_pytorch_model.safetensors"),
+    ],
+)
+state_dict = load_state_dict("models/train/FLUX.1-dev-Controlnet-Inpainting-Beta_full/epoch-0.safetensors")
+pipe.controlnet.models[0].load_state_dict(state_dict)
+
+image = pipe(
+    prompt="a cat sitting on a chair, wearing sunglasses",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/inpaint/image_1.jpg"),
+        inpaint_mask=Image.open("data/example_image_dataset/inpaint/mask.jpg"),
+        scale=0.9
+    )],
+    height=1024, width=1024,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-Controlnet-Inpainting-Beta_full.jpg")

examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Union-alpha.py

@@ -0,0 +1,31 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from diffsynth import load_state_dict
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="InstantX/FLUX.1-dev-Controlnet-Union-alpha", origin_file_pattern="diffusion_pytorch_model.safetensors"),
+    ],
+)
+state_dict = load_state_dict("models/train/FLUX.1-dev-Controlnet-Union-alpha_full/epoch-0.safetensors")
+pipe.controlnet.models[0].load_state_dict(state_dict)
+
+image = pipe(
+    prompt="a dog",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/canny/image_1.jpg"),
+        scale=0.9,
+        processor_id="canny",
+    )],
+    height=768, width=768,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-Controlnet-Union-alpha_full.jpg")

examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Upscaler.py

@@ -0,0 +1,30 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from diffsynth import load_state_dict
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="jasperai/Flux.1-dev-Controlnet-Upscaler", origin_file_pattern="diffusion_pytorch_model.safetensors"),
+    ],
+)
+state_dict = load_state_dict("models/train/FLUX.1-dev-Controlnet-Upscaler_full/epoch-0.safetensors")
+pipe.controlnet.models[0].load_state_dict(state_dict)
+
+image = pipe(
+    prompt="a dog",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/upscale/image_1.jpg"),
+        scale=0.9
+    )],
+    height=768, width=768,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-Controlnet-Upscaler_full.jpg")

examples/flux/model_training/validate_full/FLUX.1-dev-InfiniteYou.py

@@ -0,0 +1,33 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from diffsynth import load_state_dict
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="ByteDance/InfiniteYou", origin_file_pattern="infu_flux_v1.0/aes_stage2/image_proj_model.bin"),
+        ModelConfig(model_id="ByteDance/InfiniteYou", origin_file_pattern="infu_flux_v1.0/aes_stage2/InfuseNetModel/*.safetensors"),
+    ],
+)
+state_dict = load_state_dict("models/train/FLUX.1-dev-InfiniteYou_full/epoch-0.safetensors")
+state_dict_projector = {i.replace("image_proj_model.", ""): state_dict[i] for i in state_dict if i.startswith("image_proj_model.")}
+pipe.image_proj_model.load_state_dict(state_dict_projector)
+state_dict_controlnet = {i.replace("controlnet.models.0.", ""): state_dict[i] for i in state_dict if i.startswith("controlnet.models.0.")}
+pipe.controlnet.models[0].load_state_dict(state_dict_controlnet)
+
+image = pipe(
+    prompt="a man with a red hat",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/infiniteyou/image_1.jpg"),
+    )],
+    height=1024, width=1024,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-InfiniteYou_full.jpg")

examples/flux/model_training/validate_full/Step1X-Edit.py

@@ -0,0 +1,25 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig
+from diffsynth import load_state_dict
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="Qwen/Qwen2.5-VL-7B-Instruct"),
+        ModelConfig(model_id="stepfun-ai/Step1X-Edit", origin_file_pattern="step1x-edit-i1258.safetensors"),
+        ModelConfig(model_id="stepfun-ai/Step1X-Edit", origin_file_pattern="vae.safetensors"),
+    ],
+)
+state_dict = load_state_dict("models/train/Step1X-Edit_full/epoch-0.safetensors")
+pipe.dit.load_state_dict(state_dict)
+
+image = pipe(
+    prompt="Make the dog turn its head around.",
+    step1x_reference_image=Image.open("data/example_image_dataset/2.jpg").resize((768, 768)),
+    height=768, width=768, cfg_scale=6,
+    seed=0
+)
+image.save("image_Step1X-Edit_full.jpg")

examples/flux/model_training/validate_lora/FLEX.2-preview.py

@@ -0,0 +1,18 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="ostris/Flex.2-preview", origin_file_pattern="Flex.2-preview.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+    ],
+)
+pipe.load_lora(pipe.dit, "models/train/FLEX.2-preview_lora/epoch-4.safetensors", alpha=1)
+
+image = pipe(prompt="dog,white and brown dog, sitting on wall, under pink flowers", seed=0)
+image.save("image_FLEX.2-preview_lora.jpg")

examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Inpainting-Beta.py

@@ -0,0 +1,29 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta", origin_file_pattern="diffusion_pytorch_model.safetensors"),
+    ],
+)
+pipe.load_lora(pipe.dit, "models/train/FLUX.1-dev-Controlnet-Inpainting-Beta_lora/epoch-4.safetensors", alpha=1)
+
+image = pipe(
+    prompt="a cat sitting on a chair, wearing sunglasses",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/inpaint/image_1.jpg"),
+        inpaint_mask=Image.open("data/example_image_dataset/inpaint/mask.jpg"),
+        scale=0.9
+    )],
+    height=1024, width=1024,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-Controlnet-Inpainting-Beta_lora.jpg")

examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Union-alpha.py

@@ -0,0 +1,29 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="InstantX/FLUX.1-dev-Controlnet-Union-alpha", origin_file_pattern="diffusion_pytorch_model.safetensors"),
+    ],
+)
+pipe.load_lora(pipe.dit, "models/train/FLUX.1-dev-Controlnet-Union-alpha_lora/epoch-4.safetensors", alpha=1)
+
+image = pipe(
+    prompt="a dog",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/canny/image_1.jpg"),
+        scale=0.9,
+        processor_id="canny",
+    )],
+    height=768, width=768,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-Controlnet-Union-alpha_lora.jpg")

examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Upscaler.py

@@ -0,0 +1,28 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="jasperai/Flux.1-dev-Controlnet-Upscaler", origin_file_pattern="diffusion_pytorch_model.safetensors"),
+    ],
+)
+pipe.load_lora(pipe.dit, "models/train/FLUX.1-dev-Controlnet-Upscaler_lora/epoch-4.safetensors", alpha=1)
+
+image = pipe(
+    prompt="a dog",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/upscale/image_1.jpg"),
+        scale=0.9
+    )],
+    height=768, width=768,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-Controlnet-Upscaler_lora.jpg")

examples/flux/model_training/validate_lora/FLUX.1-dev-IP-Adapter.py

@@ -0,0 +1,26 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="InstantX/FLUX.1-dev-IP-Adapter", origin_file_pattern="ip-adapter.bin"),
+        ModelConfig(model_id="google/siglip-so400m-patch14-384"),
+    ],
+)
+pipe.load_lora(pipe.dit, "models/train/FLUX.1-dev-IP-Adapter_lora/epoch-4.safetensors", alpha=1)
+
+image = pipe(
+    prompt="dog,white and brown dog, sitting on wall, under pink flowers",
+    ipadapter_images=Image.open("data/example_image_dataset/1.jpg"),
+    height=768, width=768,
+    seed=0
+)
+image.save("image_FLUX.1-dev-IP-Adapter_lora.jpg")

examples/flux/model_training/validate_lora/FLUX.1-dev-InfiniteYou.py

@@ -0,0 +1,28 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig, ControlNetInput
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="flux1-dev.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder/model.safetensors"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="text_encoder_2/"),
+        ModelConfig(model_id="black-forest-labs/FLUX.1-dev", origin_file_pattern="ae.safetensors"),
+        ModelConfig(model_id="ByteDance/InfiniteYou", origin_file_pattern="infu_flux_v1.0/aes_stage2/image_proj_model.bin"),
+        ModelConfig(model_id="ByteDance/InfiniteYou", origin_file_pattern="infu_flux_v1.0/aes_stage2/InfuseNetModel/*.safetensors"),
+    ],
+)
+pipe.load_lora(pipe.dit, "models/train/FLUX.1-dev-InfiniteYou_lora/epoch-4.safetensors", alpha=1)
+
+image = pipe(
+    prompt="a man with a red hat",
+    controlnet_inputs=[ControlNetInput(
+        image=Image.open("data/example_image_dataset/infiniteyou/image_1.jpg"),
+    )],
+    height=1024, width=1024,
+    seed=0, rand_device="cuda",
+)
+image.save("image_FLUX.1-dev-InfiniteYou_lora.jpg")

examples/flux/model_training/validate_lora/Step1X-Edit.py

@@ -0,0 +1,23 @@
+import torch
+from diffsynth.pipelines.flux_image_new import FluxImagePipeline, ModelConfig
+from PIL import Image
+
+
+pipe = FluxImagePipeline.from_pretrained(
+    torch_dtype=torch.bfloat16,
+    device="cuda",
+    model_configs=[
+        ModelConfig(model_id="Qwen/Qwen2.5-VL-7B-Instruct"),
+        ModelConfig(model_id="stepfun-ai/Step1X-Edit", origin_file_pattern="step1x-edit-i1258.safetensors"),
+        ModelConfig(model_id="stepfun-ai/Step1X-Edit", origin_file_pattern="vae.safetensors"),
+    ],
+)
+pipe.load_lora(pipe.dit, "models/train/Step1X-Edit_lora/epoch-4.safetensors", alpha=1)
+
+image = pipe(
+    prompt="Make the dog turn its head around.",
+    step1x_reference_image=Image.open("data/example_image_dataset/2.jpg").resize((768, 768)),
+    height=768, width=768, cfg_scale=6,
+    seed=0
+)
+image.save("image_Step1X-Edit_lora.jpg")