TinyFT: Lightweight Fine-Tuning Library

TinyFT is a lightweight, modular toy-scale library designed from scratch for fine-tuning LLMs.

View on GitHub



DP-ZeRO Trainer

Differential Privacy experimentations, and implementation of Differential Privacy training with Zero Redundancy Optimizer support.

View on GitHub



Daily Quantitative Trading backtesting.

A toy-scale CLI-based backtesting Quantitative Daily Trading system with support for various common trading strategies.

View on GitHub