Search papers, labs, and topics across Lattice.
This study introduces a low-rank decomposition adapter and a controlled dataset designed to evaluate offline satellite-security plan decomposition, featuring 24 authored decompositions and 83 derived examples. The methodology employs a leakage-controlled approach to prevent reference-plan leakage, allowing for a fair assessment of model performance across various adapter configurations and baseline models. Key findings reveal that while the adapter shows improved precision and format compliance, the overall autoregressive next-step accuracy remains low, indicating that further development is needed for reliable decomposition systems.
Adapter precision outperforms two-shot prompting by a notable margin, yet the autoregressive next-step accuracy remains surprisingly low across all model sizes.
Some security-development settings require local models that map an objective to an ordered, checkable plan. We present a low-rank decomposition adapter and release a case-disjoint corpus with 24 authored decompositions and 83 derived next-step examples across 24 satellite-security cases. To prevent reference-plan leakage, prompts contain one objective and eight shuffled SPARTA candidates. The evaluation candidate sets are constructed with oracle inclusion of every reference technique, so this study measures selection from a controlled set, not retrieval. On six fixed cases and five decoding seeds, mean scores vary across adapters and prompted Qwen2.5 baselines; a separate greedy check over three adapter training seeds shows non-negligible training variance. A 24-fold greedy leave-one-case-out control at the two compact sizes likewise shows tradeoffs: at 1.5B, adapter precision is 0.583 against 0.480 for two-shot prompting, while recall is 0.586 against 0.660. Case-resampling intervals for adapter precision relative to the strongest prompted baseline cross zero at 1.5B and 7B. The adapter is also substantially more format-compliant, which prevents attributing all score differences to selection. Autoregressive next-step accuracy is 0.08, 0.06, and 0.29 at 0.5B, 1.5B, and 7B. We therefore offer the formulation, leakage-controlled dataset, reproducibility package, and descriptive analysis as a proof of concept, not evidence of a reliable stand-alone decomposition system or a general adapter advantage. A separate diagnostic tests contract reuse.
