Repository navigation
Bump datasets from 2.21.0 to 5.0.1 in /prepare/benchmarks/torr - #1986
Merged
yoavkatz merged 1 commit intoOct 5, 2026
Merged
Conversation
dependabot
Bot
added
dependencies
Bumps [datasets](https://github.com/huggingface/datasets) from 2.21.0 to 5.0.1. - [Release notes](https://github.com/huggingface/datasets/releases) - [Commits](huggingface/datasets@2.21.0...5.0.1) --- updated-dependencies: - dependency-name: datasets dependency-version: 5.0.1 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com>
dependabot
Bot
force-pushed
the
dependabot/pip/prepare/benchmarks/torr/datasets-5.0.1
branch
from
October 4, 2026 11:00
608a3d5 to
919a1a0
Compare
yoavkatz
added a commit
that referenced
this pull request
Oct 4, 2026
GitHub withholds repository secrets from workflow runs triggered by forks and by Dependabot. Two steps treated those secrets as mandatory and failed hard when they resolved to an empty string, which is why PR #1986 showed 11 red jobs that never executed a single test. The internal-dependency step wrote secrets.LLMEVALKIT_SSH_KEY to ~/.ssh/id_ed25519 and pip-installed from git+ssh://github.ibm.com. With an empty key this aborts the whole job: Load key "/home/runner/.ssh/id_ed25519": error in libcrypto git@github.ibm.com: Permission denied (publickey). That killed unittests, eager, and all ten preparation shards before the test step. The host is IBM-internal, so it is unreachable from an external fork regardless of the secret. Skip the install when no key is present. The llmevalkit imports in metrics.py are already lazy and guarded by _requirements_list, so the affected metrics raise a clear "install with..." message if something actually instantiates them; everything else is unaffected. The Hugging Face login step interpolated an empty token into `huggingface-cli login --token`, looped five times, and left the run unauthenticated, which then hit the shared-runner rate limit: urllib.error.HTTPError: HTTP Error 429: Too Many Requests Skip the login when no token is present and say so in the log, instead of retrying a call that cannot succeed. Unauthenticated runs remain exposed to Hub rate limits; this reports the cause rather than fixing it. Both guards are no-ops on trusted runs. Prefer a guard inside the step over continue-on-error at the call sites, so a real install failure still fails the job when the secret IS available. Also pass both secrets through env: rather than interpolating them into the script body, and write the key with printf instead of echo, which mangles values containing backslashes or a leading dash. Signed-off-by: Yoav Katz <katz@il.ibm.com> Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
yoavkatz
deleted the
dependabot/pip/prepare/benchmarks/torr/datasets-5.0.1
branch
October 5, 2026 07:20
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bumps datasets from 2.21.0 to 5.0.1.
Release notes
Sourced from datasets's releases.
... (truncated)
Commits
921c2a7release: 5.0.1 (#8370)c6fc5cdPreserve nullable integer columns in to_json/to_csv/to_sql (#8366)6747b87Fix DatasetDict.push_to_hub leaving removed splits in the dataset card (#8367)b305031fix buckets on windows (#8369)030a3e5Decode Json() columns in Dataset.to_pandas() (#8344)0f207a0Rebatch arrow source before formatting in IterableDataset.filter to fix resum...adad35dKeep integers on the python read path for fixed-shape ArrayXD columns with nu...8966746Fix CSV loader dropping on_bad_lines/encoding_errors on pandas 2.0-2.2 (#8358)b8e861aFix bucket dataset card handling and push metadata accounting (#8354)521a590Keep flat numeric columns with nulls numeric in numpy format (#8352)