kaylode/theseus
kaylode/theseus has 36 GitHub stars and 4 forks. GitHot measured 0 stars in 24 hours, 0 in 7 days, and 0 in 15 days. Last checked 2026-11-19 04:59 UTC. Last push 2026-10-11.
实测星标增长
- 24小时
- 0
- 7天
- 0
- 15天
- 0
- 1个月
- 正在收集历史
各周期内的星标净变化,周期相互重叠,并非每日新增。缺失的测量保持空白。星标代表关注度,不代表软件质量。
关于
General template for most Pytorch projects
机器学习README.md
原文:sailboat: Ship of Theseus
The ship wherein Theseus and the youth of Athens returned from Crete had thirty oars, and was preserved by the Athenians down even to the time of Demetrius Phalereus, for they took away the old planks as they decayed, putting in new and stronger timber in their places, insomuch that this ship became a standing example among the philosophers, for the logical question of things that grow; one side holding that the ship remained the same, and the other contending that it was not the same. — Plutarch, Theseus
:pencil: Instructions
Installation
- Create virtual environment:
conda create -n myenv python=3.10 - Install Pytorch. Currently support
torch==1.13.1 - Inside your project, install this package by
git+https://github.com/kaylode/theseus.git@master#egg=theseus[cv,cv.classification,cv.detection,cv.semantic]extra packages can be identified from the pyproject.toml.
To adapt for personal project
- Create your own dataset, dataloader, model, loss function, metric function, ... and register it to the registry so that it can be generated from config at runtime.
- Customize inherited trainer and pipeline to your need, such as what to do before/after training/validating step,...
- Write custom callbacks (recommended!), follow Lightning
- Modify configuration file
See theseus/cv/classification for example
To execute scripts with arguments
- Run the script with
--config-dirflag with a specified config folder that contains the yaml file. And--config-nameis that file's name.
Example:
python train.py \
--config-dir configs \
--config-name pipeline.yaml
- To override arguments inside the .yaml file, follow the instructions from Hydra. For example, to train 50 epochs and resume training from checkpoints:
python train.py \
--config-dir configs \
--config-name pipeline.yaml \
trainer.args.max_epochs=5000 \
global.resume=checkpoint.pth
Notice: There are no spaces between keys and values in -o flag
:school_satchel: Resources
- Example colab notebooks for classification tasks:
- Example colab notebooks for semantic segmentation tasks:
:blue_book: References
- This repo is inspired by https://github.com/vltanh/torchan Remember to give it a star .
- The big refactor is mostly adapted from https://github.com/HCMUS-ROBOTICS/ssdf-nncore 's nncore, which also deserves stars.