EscherNet运行笔记

文章标题：EscherNet: A Generative Model for Scalable View Synthesis

1. 环境配置

conda env create -f environment.yml -n eschernet
conda activate eschernet

2. 数据下载

wget https://tri-ml-public.s3.amazonaws.com/datasets/views_release.tar.gz

3. 推理

bash eval_eschernet.sh

4. 训练

accelerate launch train_eschernet.py --train_data_dir /data/objectverse/views_release --pretrained_model_name_or_path runwayml/stable-diffusion-v1-5 --train_batch_size 256 --dataloader_num_workers 16 --mixed_precision bf16 --gradient_checkpointing --T_in 3 --T_out 3 --T_in_val 10 --output_dir logs_N3M3B256_SD1.5

参考文献

GitHub - kxhit/EscherNet: [CVPR2024 Oral] EscherNet: A Generative Model for Scalable View Synthesis