Skip to main content

更方便的深度学习

Project description

byzh-ai

更方便的深度学习。

pip install byzh-ai → 导入名 byzh_ai

模块一览

所有子包以 B 开头,类以 B_ 开头,函数以 b_ 开头。

模块 说明
Bdata 主流数据集一键下载/缓存/标准化(MNIST、CIFAR、FashionMNIST、Tiny ImageNet 等),支持 TensorDataset / DataLoader 构建与分层采样
Bdemo MNIST 完整训练演示、训练器功能测试(单卡/多卡)、分类评估测试
Bearly_stop 多种早停策略(精度停滞、精度增量不足、损失停滞、过拟合检测)及模型参数自动回滚重载
Blr_schedulers 自定义学习率调度器:Warmup 线性升温、阶梯衰减、Warmup + 衰减组合
Bmodel 即用型深度学习模型:SimpleCNN、LeNet-5、ResNet-18、RNN、LSTM、Transformer、脉冲神经网络(Spike RNN/LSTM/ResNet-18)等
Btrainer 通用训练器,封装训练/验证/测试循环,集成早停/重载、学习率调度、checkpoint、多 GPU(DataParallel / DDP)、SNN 支持、日志记录、损失/精度绘图
Butils 深度学习工具:自动设备选择(NPU/GPU/CPU)、GPU 监控与空闲检测、模型参数量/FLOPs 统计、随机种子统一设置、数据集均值/标准差计算、超参数网格搜索、递增运行目录管理
Bvisual 训练损失/精度曲线绘制、混淆矩阵热力图绘制

依赖

  • byzh-core>=0.0.10.0 — 进度条、日志写入器、彩色终端输出、文件/目录操作等基础工具
  • torchtorchvision — 深度学习框架
  • numpypandasmatplotlibseaborn — 数据处理与可视化
  • scikit-learn — 数据划分、指标计算
  • thop — FLOPs 计算(需自行安装)
  • spikingjelly — 脉冲神经网络(可选)
  • PyYAML — YAML 配置解析
  • Pillow — 图像处理

License

MIT

Project details


Release history Release notifications | RSS feed

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

byzh_ai-0.0.10.12.tar.gz (111.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

byzh_ai-0.0.10.12-py3-none-any.whl (187.0 kB view details)

Uploaded Python 3

File details

Details for the file byzh_ai-0.0.10.12.tar.gz.

File metadata

  • Download URL: byzh_ai-0.0.10.12.tar.gz
  • Upload date:
  • Size: 111.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.5

File hashes

Hashes for byzh_ai-0.0.10.12.tar.gz
Algorithm Hash digest
SHA256 bac0904b2bbd9f93eef0f297d499e58283f8a595939a3147716d7b6866cce1e0
MD5 6693dfce6ab02821cc321813bde3fe0b
BLAKE2b-256 6399b03cfe5452d64b07ffb298aff02c495b418b618f5f2f309c708cef897b3c

See more details on using hashes here.

File details

Details for the file byzh_ai-0.0.10.12-py3-none-any.whl.

File metadata

  • Download URL: byzh_ai-0.0.10.12-py3-none-any.whl
  • Upload date:
  • Size: 187.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.5

File hashes

Hashes for byzh_ai-0.0.10.12-py3-none-any.whl
Algorithm Hash digest
SHA256 a894b547b8a865137178c6bd74b1603a690af5d9571a769639e5341c77125afd
MD5 2763cb99655dca7c2d2746fc4b395323
BLAKE2b-256 0e44808a2c6940a9723cad53a6b61e2e7e090687d33c5e7a4b207257250c06e4

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page