모드 분대
1.0.0
이것은 Mod-Squad: Designing Mixtures of Experts As Modular Multi-Task Learners 논문의 PyTorch/GPU 구현입니다.
@article{chen2022modsquad,
title={Mod-Squad: Designing Mixtures of Experts As Modular Multi-Task Learners},
author={Zitian Chen and Yikang Shen and Mingyu Ding and Zhenfang Chen and Hengshuang Zhao and Erik Learned-Miller and Chuang Gan},
journal={CVPR},
year={2023}
}
데이터세트: 태스크노미(Taskonomy)
작은 하위 집합에서 다운로드하는 예
omnitools.download class_object class_scene depth_euclidean depth_zbuffer edge_occlusion edge_texture keypoints2d keypoints3d nonfixated_matches normal points principal_curvature reshading rgb segment_semantic segment_unsup2d segment_unsup25d --components taskonomy --subset tiny --dest ./taskonomy_tiny/ --connections_total 40 --agree --name [your name] --email [your email]
./data에 데이터를 넣어주세요
기본 모델은 ./work_dir에 저장되고, 로그는 ./log_dir에 저장됩니다.
환경: timm==0.3.2 pytorch==1.10.2
MoE 모듈 설치:
cd parallel_linear
pip3 install .
python -m torch.distributed.launch --nnodes=1 --nproc_per_node=2 --master_port 44875 main_mt.py
--batch_size 6
--epochs 100
--input_size 224
--blr 4e-4 --weight_decay 0.05
--warmup_epochs 10
--model mtvit_taskgate_att_mlp_base_MI_twice
--drop_path 0.1
--scaleup
--exp-name scaleup_mtvit_taskgate_att_mlp_base_MI_twice