DLRover
Ant GroupDistributed deep-learning training system from Ant Group, built to keep large training jobs running. It manages elastic training on Kubernetes - scaling workers up and down during a run, detecting and replacing failed nodes, and resuming from checkpoints - so that a job survives the failures that would otherwise restart it.
The GitHub API reports NOASSERTION because the LICENSE opens with a preamble before the Apache text; the body reads 'The following Apache license applies to all files unless otherwise noted'.
Openness
5 high confidence- license
- Apache-2.0(OSI
- source
- public(intelligent-machine-learning/dlrover is the product you run)
- core features withheld
- no — a 1,514-entry recursive tree carries no path matching enterprise, ee, commercial or proprietary
DLRover is Apache-2.0, read from the LICENSE body rather than the GitHub classifier label. The repository tree carries no enterprise, ee or commercial path.
- https://api.github.com/repos/intelligent-machine-learning/dlrover/git/trees/master?recursive=1 recorded 2026-09-16
Full untruncated recursive tree of the default branch, 1,514 entries. Licenses are the root LICENSE and vendored dependencies. No path matches enterprise, ee, commercial or proprietary. A tree lists paths; it is cited for that and for nothing about the vendor's offerings.
- https://github.com/intelligent-machine-learning/dlrover recorded 2026-09-16
The dlrover repository page. The repository carries the system itself and names no commercial edition; cited for the license, the public source, and for there being no paid tier declared anywhere the project publishes.
- https://raw.githubusercontent.com/intelligent-machine-learning/dlrover/master/LICENSE recorded 2026-09-17
The LICENSE body, read rather than the classifier label. It opens "The following Apache license applies to all files unless otherwise noted:" and then carries the verbatim Apache License 2.0 text. That preamble line is what GitHub's classifier cannot match, and it is the whole reason the API reports NOASSERTION.
Adoption
1 high confidencePyPI downloads of the dlrover package are the adoption signal, since the package is published from the same repository as the product. The count reflects pip installs, not how many clusters actually run DLRover.
- https://pypistats.org/api/packages/dlrover/recent recorded 2026-09-16
Trailing-window download counts for the declared PyPI artifact: 295 downloads in the trailing month.
Capability
3 medium confidenceDLRover treats a training run as a first-class job that it schedules, tracks and repairs: it decides when workers start, stop and restart, but leaves pod placement to the Kubernetes scheduler. Kubeflow Trainer does the same for the same substrate, so the two sit level.
- https://github.com/intelligent-machine-learning/dlrover recorded 2026-09-16
Repository page and README for dlrover, describing elastic distributed training on Kubernetes with automatic scaling, fault detection and resume.
Verified 2026-09-16