PhoebusSi/Alpaca-CoT
An instruction-tuning platform unifying interfaces for LLMs, instruction datasets, and parameter-efficient fine-tuning methods.
Velocity · 7d
+2.4
★ / day
Trend
→steady
star history
The project provides a unified interface for instruction-tuning large language models, handling multiple LLMs (Alpaca, LLaMA, ChatGLM, MOSS) and various parameter-efficient methods including LoRA and P-tuning. It centralizes instruction-tuning datasets with support for chain-of-thought data and aims to lower the barrier for researchers to fine-tune LLMs.