Last active
January 26, 2025 23:56
-
-
Save qpwo/dd9d2e3bd5e793f342198eb28f4448b7 to your computer and use it in GitHub Desktop.
fust download with r2 rclone and huggingface parallel
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
curl https://rclone.org/install.sh | sudo bash | |
mkdir ~/.config/rclone | |
echo " | |
[r2] | |
type = s3 | |
provider = Cloudflare | |
access_key_id = $a | |
secret_access_key = $b | |
region = auto | |
endpoint = https://$c.r2.cloudflarestorage.com | |
acl = private | |
" > ~/.config/rclone/rclone.conf | |
pip install hf_transfer 'huggingface_hub[hf_transfer]' | |
export HF_HUB_ENABLE_HF_TRANSFER=1 | |
huggingface-cli login --token $d | |
started=$SECONDS | |
for digit in $(seq 0 4); do | |
huggingface-cli download --max-workers=1 --include="*${digit}-of-*.safetensors" --local-dir ~/dsv3 deepseek-ai/DeepSeek-V3 | cat & | |
rclone copy --include "*$((digit + 5))-of-*" --transfers=1 --multi-thread-streams=32 r2:hfbuk/hff/deepseek-v3-base/ ~/dsv3/ --size-only --ignore-errors --ignore-checksum --log-level=INFO | cat & | |
sleep 3 | |
done | |
wait | |
# remaining files besides .safetensors | |
rclone copy --transfers=32 r2:hfbuk/hff/deepseek-v3-base/ ~/dsv3/ --size-only --ignore-errors --ignore-checksum --log-level=INFO | cat & | |
elapsed=$((SECONDS - started)) | |
echo "took $elapsed seconds" |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment