Automated Evolutionary Hyperparameter Tuning for NLP-Based Test Case Generation

Automated generation of executable test suites from natural-language requirements remains challenging due to linguistic ambiguity and sensitivity of generative models to decoding and training hyperparameters. This paper introduces a hierarchical, multi-level evolutionary framework that treats model hyperparameters and decoding strategies as upper-level decision variables and employs lower-level fitnesses that directly measure test-quality objectives (structural coverage, semantic diversity, redundancy, and runtime efficiency). The approach integrates retrieval-augmented grounding, surrogate-assisted preselection, lightweight LoRA adaptation and optional HIL evaluation. Empirical evaluation on PURE, PROMISE_exp and FR_NFR benchmarks (repeated runs, n = 10; paired twosided t-tests, $\alpha=0.05$) shows consistent gains: on PURE mean code coverage reaches $\mathbf{8 2. 4 \%}$ (vs. $\mathbf{7 5. 1 \%}$ for Bayesian optimisation and 68.9% for random search) with 145 unique scenarios and modest runtime overhead $(\approx 58.3 \mathrm{~s}$, $\approx 6 \%$ above Bayesian). Ablations confirm component effects (e.g., removing diversity reduces unique scenarios $\approx 18 \%$; disabling the surrogate increases wall-clock $\approx \mathbf{4 2} \boldsymbol{\%}$; disabling RAG drops grounded consistency $\approx 12 \%$). Results indicate that co-optimising hyperparameters for explicit test-quality metrics, together with grounding and realistic execution, yields more useful, executable test suites. Future work will explore adaptive objective weighting, transfer warm-starts and probabilistic surrogates.