wehub-resource-sync
917eedffcf
Main / Python 3.11 - Docs (push) Has been cancelled
Main / Python 3.11 - Build (push) Has been cancelled
Main / Python 3.11 - Lint (push) Has been cancelled
Main / Python 3.11 - Style (push) Has been cancelled
Main / Python 3.11 - Test (push) Has been cancelled
Main / GPU CI (push) Has been cancelled
Main / Release (push) Has been cancelled
Main / Build and Push Docker Images (push) Has been cancelled
12 lines
339 B
Bash
Executable File
12 lines
339 B
Bash
Executable File
#!/bin/bash
|
|
|
|
datasets=(
|
|
"s2orcforolmo-unspecified-oa"
|
|
)
|
|
|
|
for dataset in "${datasets[@]}"; do
|
|
python -m olmocr.pipeline "s3://ai2-oe-data/jakep/dolma4pdfs_workspaces/${dataset}" \
|
|
--pdfs "s3://ai2-oe-data/jakep/dolma4pdfs_frontier/${dataset}/*.gz" \
|
|
--workers 2 --beaker --beaker_gpus 64 --beaker_priority high
|
|
done
|