RepoJournal
Alignment & Post-training Alignment & Post-training
47 wires and counting

$ follow Alignment & Post-training

Keep up with Alignment & Post-training in about 3 minutes: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

or

fair warning: these emails are deeply technical. diffs, version numbers, CVEs, benchmark deltas. if that's not your idea of a good read, this isn't your newsletter.

Folds into your digest — weekly by default, monthly if you prefer. Unsubscribe in one click.

$ status

wire 2026-11-17
stories 8

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

verl drops DataProto auto padding, adds single-rank shard dispatch

By RepoJournal · Filed · About Alignment & Post-training · Composed from the cited sources · methodology

verl shipped two breaking changes to how batches and shards move between ranks, plus a ROCm dataloader workaround that stops PPO steps from hanging.

[BREAKING][single_controller, data, doc] refactor: remove DataProto auto padding verl-project/verl

by vermouth1992

verl removed DataProto auto padding: `DP_COMPUTE_PROTO` now splits via `_split_args_kwargs_data_proto` and requires `len(batch) % world_size == 0`. The PR's own reasoning is that the old undo step is valid only when every worker returns the same rows in the same order at the same length, so any job relying on automatic divisibility padding must now pad explicitly with `pad_dataproto_to_divisor`.

[BREAKING][single_controller, worker, trainer] feat: dispatch shards only to the model-parallel source rank (#8708) verl-project/verl

by Chi Zhang

N-D data-parallel dispatch gains an optional path that sends each shard to exactly one model-parallel source rank and leaves redistribution to a worker-side preprocess hook, instead of copying every shard to every rank in the model-parallel group. It is breaking and the worker does not create a model-parallel process group, so anything that assumed the old per-rank copy needs the hook.

[ci, hardware] fix: set KMP_INIT_AT_FORK=TRUE on ROCm to stop dataloader fork deadlock (#8713) verl-project/verl

by Yuhan Yang

The ROCm e2e workflow and `docker/rocm/Dockerfile.rocm` now set `KMP_INIT_AT_FORK=TRUE`, which stops PPO training steps from hanging forever when dataloader workers start. Setting it in the Dockerfile means ROCm images you build from it carry the fix; the cause was `scikit-learn` and `joblib` calling `os.environ.setdefault("KMP_INIT_AT_FORK", "FALSE")` when `transformers` imports them.

Quick answers

What shipped in Alignment & Post-training on November 17, 2026?
verl shipped two breaking changes to how batches and shards move between ranks, plus a ROCm dataloader workaround that stops PPO steps from hanging. In total, 4 commits and 4 pull requests landed.
Who contributed to Alignment & Post-training on November 17, 2026?
2 developers shipped this update, including vermouth1992 and Yuhan Yang.
What were the notable Alignment & Post-training updates?
[BREAKING][single_controller, data, doc] refactor: remove DataProto auto padding, [BREAKING][single_controller, worker, trainer] feat: dispatch shards only to the model-parallel source rank (#8708), and [ci, hardware] fix: set KMP_INIT_AT_FORK=TRUE on ROCm to stop dataloader fork deadlock (#8713).