命令列環境變數設定：
https://www.cnblogs.com/fuhai0815/p/9753585.html
複製：
https://blog.csdn.net/u012964600/article/details/134926880
解壓：
https://blog.csdn.net/imyang2007/article/details/7634470
tar：
https://blog.csdn.net/qq_43657810/article/details/132328941
ubuntu版本檢視：
https://blog.csdn.net/whbing1471/article/details/52074390
vim多行刪除：
https://blog.csdn.net/xueqinmax/article/details/100153229
xz命令？：（誤區之一開始以為要下載xz，然後還想要下載yum，而且還卡住了，但是實際上沒有必要 https://computingforgeeks.com/how-to-extract-xz-files-on-linux/）
https://cn.linux-console.net/?p=3682
檔案檢視：
https://www.cnblogs.com/LovelyAmy/articles/7986959.html
scp命令：
https://blog.csdn.net/a545812327/article/details/111313810
https://blog.csdn.net/qq_29307291/article/details/72819802
conda:
路徑配置：
vim ~./bashrc
export
source ~/.bashrc
首次activate：https://blog.csdn.net/sdnuwjw/article/details/112448792
huggingface資料下載：
模型：https://blog.csdn.net/lanlinjnc/article/details/136709225
資料集：https://blog.csdn.net/sinat_29950703/article/details/143063793
載入（load_dataset）：
原理：https://developer.baidu.com/article/details/2798734
用法：https://blog.csdn.net/orangerfun/article/details/131927248
& https://www.cnblogs.com/zhangxuegold/p/17531896.html
本地載入資料集的兩個例子：
load_dataset("/defaultShare/archive/zhangyang/cache/huggingface/datasets/wikitext/wikitext-2-raw-v1", split='train') traindata = load_dataset( 'json', data_files={'train': '/defaultShare/archive/zhangyang/cache/huggingface/hub/datasets--allenai--c4/snapshots/1588ec454efa1a09f29cd18ddd04fe05fc8653a2/en/c4-train.00000-of-01024.json.gz'}, split='train' )
load_dataset(cache_dir = "")快取路徑下 一定是datasets集合下有資料嗎？
以上示例均來自claude
快取路徑
HF_HOME
git映象（不一定有效）
git clone https://hub.fastgit.org/