Answers are grounded in this paper and cite their sources.
Data, code, and analysis outputs for a comparison of retrieval-augmented generation (RAG), cache-augmented generation (CAG), and long-context prompting (LC) with 4-bit (NF4) quantized Qwen2.5-7B-Instruct on question answering over 17 Indonesian higher-education regulations (105 single-article questions; reference answers are verbatim quotations verified by script; no human raters or LLM judge). The archive contains the corpus, the QA set, raw outputs and per-item metrics of the main run and of a protocol-equalized rerun, analysis scripts, the system prompt, and environment versions. Greedy decoding; NVIDIA Tesla T4 (16 GB). This archive is not a preregistration.
No takes yet. Share an insight, caveat, or question.
Iwan Setiawan (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: