Research record
Abstract / overview
Qkabrine-automl treats circuit architecture, encoding, model type and training hyperparameters as one jointly searchable configuration space rather than separately tuned stages. The framework evaluates candidates through a consistent harness and supports multiple search strategies, while optional DQFIM estimation and barren-plateau monitoring help identify poorly trainable circuits before full training.\n\nThe paper positions qkabrine-automl within the existing AutoQML literature as a lightweight, PennyLane-native contribution with integrated trainability diagnostics, post-search circuit simplification and OpenQASM export. Its reported experiment is intentionally presented as a small reproducible illustrative run rather than evidence of performance superiority over classical baselines or competing AutoQML systems.
Publication details
- Author
- Eric Jagwara
- Affiliation
- Solid Elf Labs, Uganda
- ORCID
- 0009-0003-4935-3667
Scope & availability
This is a published research resource. The publication link points to the formal external record, where the current manuscript version and citation details can be checked directly.
Documentation expectations
Purpose, source provenance, collection or evaluation methods, quality controls, assumptions and limitations should be explicit. Any sensitive information requires an appropriate access and disclosure review before sharing.
A paper or report should distinguish evidence from interpretation, describe the analytical or evaluation method and make conclusions traceable to the underlying sources and limitations.
Related work
This publication connects directly to the Qkabrine product and the Solid Elf Qkabrine product profile.