Closed striveAgain closed 3 months ago
Hello, the param use_image_cross_attention
is set to True
by default.
I have deleted the redundant param use_ram_encoder
, please refer to the latest version.
Thanks for your remind!
OK, I have updated the latest code. Thank you!
Hello, thank you for sharing the code of SeeSR! When I read it, I found it did not seem to perform cross attention between the 'ram_encoder_hidden_states' and the resnet output during training. The screen shot is as follows, please give me some advice.
From your command
CUDA_VISIBLE_DEVICES="0,1,2,3,4,5,6,7," accelerate launch train_seesr.py \ --pretrained_model_name_or_path="preset/models/stable-diffusion-2-base" \ --output_dir="./experience/seesr" \ --root_folders 'preset/datasets/training_datasets' \ --ram_ft_path 'preset/models/DAPE.pth' \ --enable_xformers_memory_efficient_attention \ --mixed_precision="fp16" \ --resolution=512 \ --learning_rate=5e-5 \ --train_batch_size=2 \ --gradient_accumulation_steps=2 \ --null_text_ratio=0.5 --dataloader_num_workers=0 \ --checkpointing_steps=10000
, there's no----use_ram_encoder
.Thus,
![image](https://github.com/cswry/SeeSR/assets/105333061/471cb629-ce33-4dc6-85ef-8a5f48dce31f)
use_image_cross_attention
will beFalse
.While forwarding,
it will skip to
else
, omitting theimage_encoder_hidden_states
.Please give me some instructions, thanks!