方式一:modelee
modelee 当前主要是提供 huggingface 模型和数据集的镜像,方便国内开发者快速获取。
本站供您下载模型镜像文件(以下称 “镜像文件”)的知识产权归属于该镜像文件的开发者所有。您的下载和使用镜像文件的行为应遵守该等知识产权的约定。若违侵犯该等知识产权,在法律规定的范围内,由此带来的责任由您承担。
除第 1 条的镜像文件外,本站的其他内容,包括图片、网站架构与画面的安排、网页设计、文字、图表、代码、SDK、API、LOGO 等知识产权及其他合法权益,包括商标权、著作权与专利权等知识产权,均归我们或我们的关联方所有。未经我们或关联公司的事先书面许可,任何人不得以包括通过机器人、蜘蛛等程序或设备监视、复制、传播、展示、镜像、上载、下载等方式擅自使用本站内除第 1 条镜像文件之外的所有内容。
方式二:hf-mirror(推荐)
hf-mirror 是一个 HuggingFace 镜像站,提供 HuggingFace 模型和数据集的搜索。
hf-mirror.com,用于镜像 huggingface.co 域名。作为一个公益项目,致力于帮助国内 AI 开发者快速、稳定的下载模型、数据集。捐赠支持请看网页左下角,感谢支持!
使用 url 下载模型或者数据集时,将
huggingface.co
直接替换为hf-mirror.com
。使用浏览器或者wget -c
、curl -L
、aria2c
等命令行方式即可。下载需登录的模型需命令行添加
--header hf_***
参数,添加 HuggingFace 的token
。
更多详细用法请看《这篇教程》。
1)网页下载
在 hf-mirror.com 中搜索,并在模型主页的 Files and Version
中下载文件。
2)huggingface-cli
huggingface-cli 是 Hugging Face 官方提供的命令行工具,自带完善的下载功能。
- 安装依赖
pip install -U huggingface_hub
- 设置环境变量
- Linux
export HF_ENDPOINT=https://hf-mirror.com
- Windows Powershell
$env:HF_ENDPOINT = "https://hf-mirror.com"
建议将上面这一行写入 ~/.bashrc
。
3.1 下载模型
huggingface-cli download --resume-download jetmoe/jetmoe-8b --local-dir jetmoe/jetmoe-8b
3.2 下载数据集
huggingface-cli download --repo-type dataset --resume-download wikitext --local-dir wikitext
可以添加 --local-dir-use-symlinks False
参数禁用文件软链接,这样下载路径下所见即所得,详细解释请见上面提到的教程。
3)使用 hfd
hfd 是 hf-mirror 作者开发的 huggingface 专用下载工具,基于成熟工具 git + aria2
,可以做到稳定下载不断线。
- 安装 aria2(Ubuntu)
sudo apt update
sudo apt install aria2
- 下载 hfd
wget https://hf-mirror.com/hfd/hfd.sh
chmod a+x hfd.sh
- 设置环境变量
- Linux
export HF_ENDPOINT=https://hf-mirror.com
- Windows Powershell
$env:HF_ENDPOINT = "https://hf-mirror.com"
3.1 下载模型
# 使用 aria2 工具进行下载,4 线程
./hfd.sh jetmoe/jetmoe-8b --tool aria2c -x 4
运行结果为:
Downloading to jetmoe-8b
Testing GIT_REFS_URL: https://hf-mirror.com/jetmoe/jetmoe-8b/info/refs?service=git-upload-pack
git clone https://hf-mirror.com/jetmoe/jetmoe-8b jetmoe-8b
Cloning into 'jetmoe-8b'...
remote: Enumerating objects: 104, done.
remote: Counting objects: 100% (100/100), done.
remote: Compressing objects: 100% (100/100), done.
remote: Total 104 (delta 49), reused 0 (delta 0), pack-reused 4
Receiving objects: 100% (104/104), 1.81 MiB | 167.00 KiB/s, done.
Resolving deltas: 100% (49/49), done.
Start Downloading lfs files, bash script:
cd jetmoe-8b
aria2c --console-log-level=error -x 4 -s 4 -k 1M -c "https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00001-of-00004.safetensors" -d "." -o "model-00001-of-00004.safetensors"
aria2c --console-log-level=error -x 4 -s 4 -k 1M -c "https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00002-of-00004.safetensors" -d "." -o "model-00002-of-00004.safetensors"
aria2c --console-log-level=error -x 4 -s 4 -k 1M -c "https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00003-of-00004.safetensors" -d "." -o "model-00003-of-00004.safetensors"
aria2c --console-log-level=error -x 4 -s 4 -k 1M -c "https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00004-of-00004.safetensors" -d "." -o "model-00004-of-00004.safetensors"
aria2c --console-log-level=error -x 4 -s 4 -k 1M -c "https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/tokenizer.model" -d "." -o "tokenizer.model"
Start downloading model-00001-of-00004.safetensors.
*** Download Progress Summary as of Sat Apr 13 14:38:13 2024 ***
===========================================================================================================
[#73a50f 684MiB/4.5GiB(14%) CN:4 DL:11MiB ETA:5m47s]
FILE: ./model-00001-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:39:14 2024 ***
===========================================================================================================
[#73a50f 1.3GiB/4.5GiB(29%) CN:4 DL:11MiB ETA:4m48s]
FILE: ./model-00001-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:40:15 2024 ***
===========================================================================================================
[#73a50f 2.0GiB/4.5GiB(44%) CN:4 DL:11MiB ETA:3m48s]
FILE: ./model-00001-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:41:15 2024 ***
===========================================================================================================
[#73a50f 2.6GiB/4.5GiB(59%) CN:4 DL:11MiB ETA:2m47s]
FILE: ./model-00001-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:42:16 2024 ***
===========================================================================================================
[#73a50f 3.3GiB/4.5GiB(74%) CN:4 DL:11MiB ETA:1m46s]
FILE: ./model-00001-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:43:17 2024 ***
===========================================================================================================
[#73a50f 4.0GiB/4.5GiB(88%) CN:4 DL:11MiB ETA:45s]
FILE: ./model-00001-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
[#73a50f 4.5GiB/4.5GiB(99%) CN:1 DL:10MiB]
Download Results:
gid |stat|avg speed |path/URI
======+====+===========+=======================================================
73a50f|OK | 11MiB/s|./model-00001-of-00004.safetensors
Status Legend:
(OK):download completed.
Downloaded https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00001-of-00004.safetensors successfully.
Start downloading model-00002-of-00004.safetensors.
*** Download Progress Summary as of Sat Apr 13 14:45:04 2024 ***
===========================================================================================================
[#bef033 668MiB/4.5GiB(14%) CN:4 DL:11MiB ETA:5m56s]
FILE: ./model-00002-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:46:04 2024 ***
===========================================================================================================
[#bef033 1.3GiB/4.5GiB(28%) CN:4 DL:11MiB ETA:4m56s]
FILE: ./model-00002-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:47:05 2024 ***
===========================================================================================================
[#bef033 1.9GiB/4.5GiB(43%) CN:4 DL:11MiB ETA:3m55s]
FILE: ./model-00002-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:48:06 2024 ***
===========================================================================================================
[#bef033 2.6GiB/4.5GiB(58%) CN:4 DL:11MiB ETA:2m52s]
FILE: ./model-00002-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:49:07 2024 ***
===========================================================================================================
[#bef033 3.3GiB/4.5GiB(72%) CN:4 DL:11MiB ETA:1m53s]
FILE: ./model-00002-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:50:08 2024 ***
===========================================================================================================
[#bef033 4.0GiB/4.5GiB(87%) CN:4 DL:11MiB ETA:53s]
FILE: ./model-00002-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
[#bef033 4.5GiB/4.5GiB(99%) CN:2 DL:10MiB]
Download Results:
gid |stat|avg speed |path/URI
======+====+===========+=======================================================
bef033|OK | 11MiB/s|./model-00002-of-00004.safetensors
Status Legend:
(OK):download completed.
Downloaded https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00002-of-00004.safetensors successfully.
Start downloading model-00003-of-00004.safetensors.
*** Download Progress Summary as of Sat Apr 13 14:52:03 2024 ***
===========================================================================================================
[#aee76a 684MiB/4.5GiB(14%) CN:4 DL:11MiB ETA:5m53s]
FILE: ./model-00003-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:53:04 2024 ***
===========================================================================================================
[#aee76a 1.3GiB/4.5GiB(29%) CN:4 DL:11MiB ETA:4m53s]
FILE: ./model-00003-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:54:04 2024 ***
===========================================================================================================
[#aee76a 2.0GiB/4.5GiB(43%) CN:4 DL:11MiB ETA:3m55s]
FILE: ./model-00003-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:55:05 2024 ***
===========================================================================================================
[#aee76a 2.6GiB/4.5GiB(58%) CN:4 DL:11MiB ETA:2m50s]
FILE: ./model-00003-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:56:06 2024 ***
===========================================================================================================
[#aee76a 3.3GiB/4.5GiB(73%) CN:4 DL:11MiB ETA:1m50s]
FILE: ./model-00003-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:57:06 2024 ***
===========================================================================================================
[#aee76a 4.0GiB/4.5GiB(87%) CN:4 DL:11MiB ETA:50s]
FILE: ./model-00003-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
[#aee76a 4.5GiB/4.5GiB(99%) CN:2 DL:11MiB]
Download Results:
gid |stat|avg speed |path/URI
======+====+===========+=======================================================
aee76a|OK | 11MiB/s|./model-00003-of-00004.safetensors
Status Legend:
(OK):download completed.
Downloaded https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00003-of-00004.safetensors successfully.
Start downloading model-00004-of-00004.safetensors.
*** Download Progress Summary as of Sat Apr 13 14:58:58 2024 ***
===========================================================================================================
[#b781a0 665MiB/2.1GiB(30%) CN:4 DL:11MiB ETA:2m14s]
FILE: ./model-00004-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 14:59:59 2024 ***
===========================================================================================================
[#b781a0 1.3GiB/2.1GiB(61%) CN:4 DL:11MiB ETA:1m13s]
FILE: ./model-00004-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
*** Download Progress Summary as of Sat Apr 13 15:00:59 2024 ***
===========================================================================================================
[#b781a0 1.9GiB/2.1GiB(93%) CN:4 DL:11MiB ETA:13s]
FILE: ./model-00004-of-00004.safetensors
-----------------------------------------------------------------------------------------------------------
[#b781a0 2.1GiB/2.1GiB(99%) CN:4 DL:11MiB]
Download Results:
gid |stat|avg speed |path/URI
======+====+===========+=======================================================
b781a0|OK | 11MiB/s|./model-00004-of-00004.safetensors
Status Legend:
(OK):download completed.
Downloaded https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/model-00004-of-00004.safetensors successfully.
Start downloading tokenizer.model.
[#1bca27 0B/0B CN:1 DL:0B]
Download Results:
gid |stat|avg speed |path/URI
======+====+===========+=======================================================
1bca27|OK | 2.1MiB/s|./tokenizer.model
Status Legend:
(OK):download completed.
Downloaded https://hf-mirror.com/jetmoe/jetmoe-8b/resolve/main/tokenizer.model successfully.
Download completed successfully.
3.2 下载数据集
./hfd.sh wikitext --dataset --tool aria2c -x 4
hfd
的完整用法如下:
./hfd.sh -h
Usage:
hfd <repo_id> [--include include_pattern] [--exclude exclude_pattern] [--hf_username username] [--hf_token token] [--tool aria2c|wget] [-x threads] [--dataset] [--local-dir path]
Description:
Downloads a model or dataset from Hugging Face using the provided repo ID.
Parameters:
repo_id The Hugging Face repo ID in the format 'org/repo_name'.
--include (Optional) Flag to specify a string pattern to include files for downloading.
--exclude (Optional) Flag to specify a string pattern to exclude files from downloading.
include/exclude_pattern The pattern to match against filenames, supports wildcard characters. e.g., '--exclude *.safetensor', '--include vae/*'.
--hf_username (Optional) Hugging Face username for authentication. **NOT EMAIL**.
--hf_token (Optional) Hugging Face token for authentication.
--tool (Optional) Download tool to use. Can be aria2c (default) or wget.
-x (Optional) Number of download threads for aria2c. Defaults to 4.
--dataset (Optional) Flag to indicate downloading a dataset.
--local-dir (Optional) Local directory path where the model or dataset will be stored.
Example:
hfd bigscience/bloom-560m --exclude *.safetensors
hfd meta-llama/Llama-2-7b --hf_username myuser --hf_token mytoken -x 4
hfd lavita/medical-qa-shared-task-v1-toy --dataset
4)使用环境变量(非侵入式)
非侵入式,能解决大部分情况。huggingface 工具链会获取 HF_ENDPOINT
环境变量来确定下载文件所用的网址,所以可以使用通过设置变量来解决。
HF_ENDPOINT=https://hf-mirror.com python your_script.py
不过有些数据集有内置的下载脚本,那就需要手动改一下脚本内的地址来实现了。
常见问题
Q: 有些项目需要登录,如何下载?
A:部分 Gated Repo 需登录申请许可。为保障账号安全,本站不支持登录,需先前往 Hugging Face 官网登录、申请许可,在官网这里获取 Access Token 后回镜像站用命令行下载。
部分工具下载 Gated Repo 的方法:
- huggingface-cli: 添加
--token
参数
huggingface-cli download --token hf_*** --resume-download meta-llama/Llama-2-7b-hf --local-dir Llama-2-7b-hf
- hfd: 添加
--hf_username
、--hf_token
参数
hfd meta-llama/Llama-2-7b --hf_username YOUR_HF_USERNAME --hf_token hf_***
其余如 from_pretrained
、wget
、curl
如何设置认证 token,详见上面第一段提到的教程。
总结
如果各位有魔法,可以直接使用魔法来访问、下载 HuggingFace 的模型和数据集。
如果没有魔法,可以尝试使用 HuggingFace 镜像进行下载。
如果需要在 Linux 中尝试使用魔法,可以参考 Clash-for-Linux 的安装与使用
参考
文档信息
- 本文作者:Bookstall
- 本文链接:https://bookstall.github.io/fragment/2023-12-16-huggingface-mirror/
- 版权声明:自由转载-非商用-非衍生-保持署名(创意共享3.0许可证)