Datasets:

Languages:
English
Size:
n>1T
ArXiv:
License:

How long time it takes to download or load the whole dataset?

#45
by yliu2702 - opened

I've failed several times to download or load the dataset. What's the estimated time for the whole process? Thanks a lot! Also, the downloading file types is json.gz, I wonder if I load the dataset, do I need to unzip it? Thanks!!

I downloaded it locally -- It took me about a day (24 hours). It very much depends on your web connection.
I also used parallelism.
You can use my update of the download script:

DATA_DIR="<path/to/your/download/folder>"
PARALLEL_DOWNLOADS="4"
DOLMA_VERSION="v1_7"

git clone https://huggingface.co/datasets/allenai/dolma
mkdir -p "${DATA_DIR}"


# cat "dolma/urls/${DOLMA_VERSION}.txt" | xargs -n 1 -P "${PARALLEL_DOWNLOADS}" wget -q -P "$DATA_DIR" # commented out the original command, leaving it for a reference
cat "dolma/urls/${DOLMA_VERSION}.txt" | xargs -n 1 -P "${PARALLEL_DOWNLOADS}" wget --show-progress --progress=bar:force:noscroll -x -ncH --cut-dirs=1 -P "$DATA_DIR"

Sign up or log in to comment