ses-conda 2.1.0

A simple deeplearning kit, implementing the module concept of SES Conda

SES Conda 2.1.0 自述文件/README

中英双语 | Bilingual
SES Conda是一个轻量深度学习套件，利用EricasZ的思路，实现简单的数据推断。
SES Conda is a lightweight deeplearning toolkit, implementing simple data prediction following the idea of EricasZ

---

1. 30 秒上手 / 30-Second Start

# 深度学习回归 / Deep regression
from sesconda.dlkit import RegressionKit
X, y = np.loadtxt('X.txt'), np.loadtxt('y.txt')
model_fn = RegressionKit('numpy', X_path='X.npy', y_path='y.npy').run()
print(model_fn([[0.5, 1.2]]))

# 经典多项式拟合 / Classic polynomial fit
from sesconda import expo
e = expo()
e.train([(0, 1), (1, 2.7), (2, 7.4)])
e.display()

---

2. 功能总览 / Features Overview

模块 / Module	能力 / Capability	适用场景 / Use-Case
RegressionKit	任意非线性黑盒逼近 / Universal nonlinear approximation	大数据、未知解析式 / Big data & unknown formula
expo	最优多项式分段拟合 / Optimal polynomial segment fit	小样本、快速可视化 / Small sample & quick plot

---

3. API 速查 / Quick API

3.1 RegressionKit（深度学习 / DL）

kit = RegressionKit(source, **kwargs)

数据源 / Source	示例 / Example
csv	RegressionKit('csv', path='data.csv', x_cols=[0,1], y_col=2)
numpy	RegressionKit('numpy', X_path='X.npy', y_path='y.npy')
sklearn	RegressionKit('sklearn', name='california')
function	RegressionKit('func', func=lambda x: np.sin(x))

可调关键参数 / Key args

model_fn = kit.run(
    hidden_dims=(256, 256, 128),
    epochs=800,
    patience=50,
    lr=1e-3,
    batch_size=128,
    standardize=True
)

3.2 expo（经典 / Classic）

e = expo()
e.train([(x1, y1), (x2, y2), ...])  # 支持批量 / Batch input
e.display()                           # 自动选阶 & 绘图 / Auto degree & plot

---

4. 更新日志 / Changelog

版本 / Version	日期 / Date	亮点 / Highlights
2.1.0	2025-08-13	并入 dlkit，新增 RegressionKit / Integrated dlkit, added RegressionKit
2.0.1	2025-07-xx	Bug 修复 / Bug fixes
2.0.0-dev	2025-06-xx	scikit-learn 情感分析优化 / Sentiment analysis improvement
1.x	…	早期多项式与关系解析 / Early polynomial & relation parse

---

5. 安装 / Installation

pip install -U ses-conda

依赖 / Requirements
torch ≥ 1.9, scikit-learn ≥ 1.3, numpy, matplotlib, pandas

---

6. 许可证 / License

MIT © 2024-End of the Time EricasZ & 街角的猫_wjz
详见 LICENSE.txt / See LICENSE.txt for details.

---