DiffSynth-Studio 2.0 major update

examples/qwen_image/model_training/special/differential_training/Qwen-Image-LoRA.sh

@@ -0,0 +1,40 @@
+# This script is provided as an example only.
+# Please manually replace the two datasets:
+# the first training dataset should contain content you do not want to generate,
+# and the second training dataset should contain content you do want to generate.
+
+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 "Qwen/Qwen-Image:transformer/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-LoRA-deterministic" \
+  --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 \
+  --use_gradient_checkpointing \
+  --dataset_num_workers 8 \
+  --find_unused_parameters
+
+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 "Qwen/Qwen-Image:transformer/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-LoRA-differencial" \
+  --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 \
+  --use_gradient_checkpointing \
+  --dataset_num_workers 8 \
+  --find_unused_parameters \
+  --preset_lora_path "./models/train/Qwen-Image-LoRA-deterministic/epoch-4.safetensors" \
+  --preset_lora_model "dit"

examples/qwen_image/model_training/special/fp8_training/Qwen-Image-LoRA.sh

@@ -0,0 +1,17 @@
+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 "Qwen/Qwen-Image:transformer/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_lora_fp8" \
+  --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 \
+  --use_gradient_checkpointing \
+  --dataset_num_workers 8 \
+  --find_unused_parameters \
+  --fp8_models "Qwen/Qwen-Image:transformer/diffusion_pytorch_model*.safetensors,Qwen/Qwen-Image:text_encoder/model*.safetensors,Qwen/Qwen-Image:vae/diffusion_pytorch_model.safetensors"

examples/qwen_image/model_training/special/fp8_training/validate.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="Qwen/Qwen-Image", origin_file_pattern="transformer/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_lora_fp8/epoch-4.safetensors")
+prompt = "a dog"
+image = pipe(prompt, seed=0)
+image.save("image.jpg")

examples/qwen_image/model_training/special/low_vram_training/Qwen-Image-LoRA.sh

@@ -0,0 +1,38 @@
+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 1 \
+  --model_id_with_origin_paths "Qwen/Qwen-Image:text_encoder/model*.safetensors,Qwen/Qwen-Image:vae/diffusion_pytorch_model.safetensors" \
+  --fp8_models "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-LoRA-lowvram-cache" \
+  --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 \
+  --use_gradient_checkpointing \
+  --use_gradient_checkpointing_offload \
+  --dataset_num_workers 8 \
+  --find_unused_parameters \
+  --task "sft:data_process"
+
+accelerate launch examples/qwen_image/model_training/train.py \
+  --dataset_base_path "./models/train/Qwen-Image-LoRA-lowvram-cache" \
+  --max_pixels 1048576 \
+  --dataset_repeat 50 \
+  --model_id_with_origin_paths "Qwen/Qwen-Image:transformer/diffusion_pytorch_model*.safetensors" \
+  --fp8_models "Qwen/Qwen-Image:transformer/diffusion_pytorch_model*.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/Qwen-Image-LoRA-lowvram" \
+  --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 \
+  --use_gradient_checkpointing \
+  --use_gradient_checkpointing_offload \
+  --dataset_num_workers 8 \
+  --find_unused_parameters \
+  --task "sft:train"

examples/qwen_image/model_training/special/simple/train.py

@@ -0,0 +1,76 @@
+import torch, accelerate
+from diffsynth.core import UnifiedDataset
+from diffsynth.pipelines.qwen_image import QwenImagePipeline, ModelConfig
+from diffsynth.diffusion import *
+
+class QwenImageTrainingModule(DiffusionTrainingModule):
+    def __init__(self, device):
+        super().__init__()
+        # Load the pipeline
+        self.pipe = QwenImagePipeline.from_pretrained(
+            torch_dtype=torch.bfloat16,
+            device=device,
+            model_configs=[
+                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"),
+                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/"),
+        )
+        # Switch to training mode
+        self.switch_pipe_to_training_mode(
+            self.pipe,
+            lora_base_model="dit",
+            lora_target_modules="to_q,to_k,to_v,add_q_proj,add_k_proj,add_v_proj",
+            lora_rank=32,
+        )
+
+    def forward(self, data):
+        # Preprocess
+        inputs_posi = {"prompt": data["prompt"]}
+        inputs_nega = {"negative_prompt": ""}
+        inputs_shared = {
+            # Assume you are using this pipeline for inference,
+            # please fill in the input parameters.
+            "input_image": data["image"],
+            "height": data["image"].size[1],
+            "width": data["image"].size[0],
+            # Please do not modify the following parameters
+            # unless you clearly know what this will cause.
+            "cfg_scale": 1,
+            "rand_device": self.pipe.device,
+            "use_gradient_checkpointing": True,
+            "use_gradient_checkpointing_offload": False,
+        }
+        for unit in self.pipe.units:
+            inputs_shared, inputs_posi, inputs_nega = self.pipe.unit_runner(unit, self.pipe, inputs_shared, inputs_posi, inputs_nega)
+        # Loss
+        loss = FlowMatchSFTLoss(self.pipe, **inputs_shared, **inputs_posi)
+        return loss
+
+if __name__ == "__main__":
+    accelerator = accelerate.Accelerator(
+        kwargs_handlers=[accelerate.DistributedDataParallelKwargs(find_unused_parameters=True)],
+    )
+    dataset = UnifiedDataset(
+        base_path="data/example_image_dataset",
+        metadata_path="data/example_image_dataset/metadata.csv",
+        repeat=50,
+        data_file_keys="image",
+        main_data_operator=UnifiedDataset.default_image_operator(
+            base_path="data/example_image_dataset",
+            height=512,
+            width=512,
+            height_division_factor=16,
+            width_division_factor=16,
+        )
+    )
+    model = QwenImageTrainingModule(accelerator.device)
+    model_logger = ModelLogger(
+        output_path="models/toy_model",
+        remove_prefix_in_ckpt="pipe.dit.",
+    )
+    launch_training_task(
+        accelerator, dataset, model, model_logger,
+        learning_rate=1e-5, num_epochs=1,
+    )

examples/qwen_image/model_training/special/split_training/Qwen-Image-LoRA.sh

@@ -0,0 +1,34 @@
+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 1 \
+  --model_id_with_origin_paths "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-LoRA-splited-cache" \
+  --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 \
+  --use_gradient_checkpointing \
+  --dataset_num_workers 8 \
+  --find_unused_parameters \
+  --task "sft:data_process"
+
+accelerate launch examples/qwen_image/model_training/train.py \
+  --dataset_base_path "./models/train/Qwen-Image-LoRA-splited-cache" \
+  --max_pixels 1048576 \
+  --dataset_repeat 50 \
+  --model_id_with_origin_paths "Qwen/Qwen-Image:transformer/diffusion_pytorch_model*.safetensors" \
+  --learning_rate 1e-4 \
+  --num_epochs 5 \
+  --remove_prefix_in_ckpt "pipe.dit." \
+  --output_path "./models/train/Qwen-Image-LoRA-splited" \
+  --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 \
+  --use_gradient_checkpointing \
+  --dataset_num_workers 8 \
+  --find_unused_parameters \
+  --task "sft:train"

examples/qwen_image/model_training/special/split_training/validate.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="Qwen/Qwen-Image", origin_file_pattern="transformer/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-LoRA-splited/epoch-4.safetensors")
+prompt = "a dog"
+image = pipe(prompt, seed=0)
+image.save("image.jpg")