aimagelab/multimodal-garment-designer

ValueError: cross_attention_dim must be specified for CrossAttnDownBlock2D

wly-ai-bj opened this issue · 1 comments

wly-ai-bj commented

when I run "python src/eval.py --dataset_path --batch_size --mixed_precision fp16 --output_dir --save_name --num_workers_test --sketch_cond_rate 0.2 --dataset <dresscode|vitonhd> --start_cond_rate 0.0 --test_order <paired|unpaired>", I met the error as follows:

ValueError: cross_attention_dim must be specified for CrossAttnDownBlock2D

wly-ai-bj commented

一直没有跑通，请问输入的cloth和pose，二者需要对齐吗？