leeguandong/MiniLLaMA3

llama3的迷你版本，包括了从0-1构造数据，训练tokenizer，pt，sft，dpo的全流程

MiniLLaMA3

Minillama3-＞训练tokenizer
https://blog.csdn.net/u012193416/article/details/139750493?spm=1001.2014.3001.5501

Minillama3-＞pt训练
https://blog.csdn.net/u012193416/article/details/139753422?spm=1001.2014.3001.5501

Minillama3-＞sft训练
https://blog.csdn.net/u012193416/article/details/139811461?spm=1001.2014.3001.5501

Minillama3-＞dpo训练
https://blog.csdn.net/u012193416/article/details/139824694?spm=1001.2014.3001.5501