RepoJournal
Alignment & Post-training Alignment & Post-training
11 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 2026-10-11
stories 4
RSS Archive Reviews

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

veRL merges veomni checkpoints, automates Ascend doc translation

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

wuxibin89 Geofeng12138
2 people shipped this
wuxibin89 @wuxibin89 1 cited Geofeng12138 @Geofeng12138 1 cited

veRL's checkpoint tooling now converts DCP checkpoints into veomni models, and the Ascend tutorial docs gain a weekly Chinese-to-English translation pipeline that runs against a separate private fork.

[ckpt] feat: add veomni model merger verl-project/verl

by wuxibin89

veRL can now produce a veomni model from an existing DCP checkpoint, so a training run saved in DCP format no longer has to be converted by hand outside the framework. Where the merger runs and what checkpoint layouts it accepts is the first thing to check before you swap it into a serving pipeline.

[doc, ci] feat: add Ascend tutorial doc translation pipeline verl-project/verl

by Geofeng12138

A new scheduled workflow, schedule_doc_translate_ascend.yaml, translates docs/ascend_tutorial/zh/ into docs/ascend_tutorial/en/ on a weekly cadence. The translation engine, skill prompts and .po translation memory live in a separate private fork, so the upstream repo receives only the English docs and the workflow file; anyone hosting a mirror or vendoring that directory should not expect the t...

$ ls alignment/reviews/ # the step-back reads

the week in review verl adds TPU SFT via TorchTitan and ETP checkpoint conversion
Full archive → Weekly reviews →

Quick answers

What shipped in Alignment & Post-training on October 11, 2026?
veRL's checkpoint tooling now converts DCP checkpoints into veomni models, and the Ascend tutorial docs gain a weekly Chinese-to-English translation pipeline that runs against a separate private fork. In total, 2 commits and 2 pull requests landed.
Who contributed to Alignment & Post-training on October 11, 2026?
2 developers shipped this update, including wuxibin89 and Geofeng12138.
What were the notable Alignment & Post-training updates?
[ckpt] feat: add veomni model merger and [doc, ci] feat: add Ascend tutorial doc translation pipeline.
♥

$ log Alignment & Post-training --recent

verl rejects conflicting vLLM engine_kwargs, open-instruct adds MILES OPD behind a Megatron learner 8 shipped FSDP2 CPU offload was silently corrupting parameters, and MILES gets FlashAttention 8 shipped Fused cross-entropy lands on Ascend NPUs and Megatron, checkpoint offload stops assuming CUDA 26 shipped DeepSeek MTP support fixed for vLLM and Megatron backends 10 shipped

$ log --all # elsewhere on the wire

Interpretability nnsight 0.8.0 cuts import time from 7.6s to 1.6s Agent Security Approval digests get RFC 8785 float serialization; sandbox-runtime 0.0.80 kills a 300 ms path

$ ls showcases/

Safety Evals AI Red Teaming Agent Security Interpretability

$ init your-desk

RepoJournal can write one of these for any repo you care about. Sign in with GitHub, pick your repos and sources, and your desk gets a wire like this one every morning.