theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-18 22:08:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

update lora and full train

Browse Source
This commit is contained in:
mi804
2025-08-18 19:09:19 +08:00
parent f9ce261a0e
commit 123f6dbadb
3 changed files with 7 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
diffsynth/pipelines/qwen_image.py
View File

@@ -623,11 +623,11 @@ def model_fn_qwen_image(
timestep = timestep / 1000
image = rearrange(latents, "B C (H P) (W Q) -> B (H W) (C P Q)", H=height//16, W=width//16, P=2, Q=2)
image_seq_len = image.shape[1]
if edit_latents is not None:
img_shapes += [(edit_latents.shape[0], edit_latents.shape[2]//2, edit_latents.shape[3]//2)]
edit_image = rearrange(edit_latents, "B C (H P) (W Q) -> B (H W) (C P Q)", H=edit_latents.shape[2]//2, W=edit_latents.shape[3]//2, P=2, Q=2)
image_seq_len = image.shape[1]
image = torch.cat([image, edit_image], dim=1)
image = dit.img_in(image)
@@ -660,8 +660,8 @@ def model_fn_qwen_image(
enable_fp8_attention=enable_fp8_attention,
)
if blockwise_controlnet_conditioning is not None:
image = image + blockwise_controlnet.blockwise_forward(
image=image, conditionings=blockwise_controlnet_conditioning,
image[:, :image_seq_len] = image[:, :image_seq_len] + blockwise_controlnet.blockwise_forward(
image=image[:, :image_seq_len], conditionings=blockwise_controlnet_conditioning,
controlnet_inputs=blockwise_controlnet_inputs, block_id=block_id,
progress_id=progress_id, num_inference_steps=num_inference_steps,
)

2
examples/qwen_image/model_training/full/Qwen-Image-Edit.sh
View File

@@ -1,6 +1,8 @@
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_edit.csv \
--data_file_keys "image,edit_image" \
--extra_inputs "edit_image" \
--max_pixels 1048576 \
--dataset_repeat 50 \
--model_id_with_origin_paths "Qwen/Qwen-Image-Edit:transformer/diffusion_pytorch_model*.safetensors,Qwen/Qwen-Image:text_encoder/model*.safetensors,Qwen/Qwen-Image:vae/diffusion_pytorch_model.safetensors" \

2
examples/qwen_image/model_training/lora/Qwen-Image-Edit.sh
View File

@@ -1,6 +1,8 @@
accelerate launch examples/qwen_image/model_training/train.py \
--dataset_base_path data/example_image_dataset \
--dataset_metadata_path data/example_image_dataset/metadata_edit.csv \
--data_file_keys "image,edit_image" \
--extra_inputs "edit_image" \
--max_pixels 1048576 \
--dataset_repeat 50 \
--model_id_with_origin_paths "Qwen/Qwen-Image-Edit:transformer/diffusion_pytorch_model*.safetensors,Qwen/Qwen-Image:text_encoder/model*.safetensors,Qwen/Qwen-Image:vae/diffusion_pytorch_model.safetensors" \