Style Without Substance: Small-Model SFT Distillation of CTI-Style Prose and What It Reveals About Detector and Safety-Classifier Calibration (Paper 2 artefacts)

This deposit accompanies the preprint "Style Without Substance: Small-Model SFT Distillation of CTI-Style Prose and What It Reveals About Detector and Safety-Classifier Calibration" (Paper 2). It contains the paper source, the eval harness, the six-axis attacker-persona rubric, per-record judgments under three Opus passes (original N=90, re-judged 15 lost records, and a fresh N=194 wider eval), a Sonnet 4.6 inter-rater pass on N=90, a Gemma 4 E4B inter-rater pass on N=15, an ATT&CK-validation output, a detector-baseline under three classifiers, and the 4,010-chunk public-CTI training corpus. v2 corrects a v1 methodological error. The v1 deposit (10.5281/zenodo.20365823) reported a 53% Opus-refusal rate on the ttp_summary transform class. Re-investigation on 2026-05-24 established that 15 of the 16 originally-reported refusals were upstream BadRequestError: credit balance too low events that returned before the request reached Opus; only 1/30 was a genuine refusal (chunk_idx=2 ttp_summary; reproduces at 1/67 on the new N=194 cut, same record both times). v2 includes the corrected eval data, a new N=194 wider eval that tightens per-axis CIs from ±0.13-0.29 to ±0.086-0.170, and the v3-vs-v4 audit trail for the companion safety-note. Headline numbers reproducible from this v2 deposit: Per-axis Opus 4.7 stylistic-fidelity means on N=194 wider eval: overall 2.76/5, style_consistency 3.46 (only axis above acceptable) down to tradecraft_vocabulary 2.33. CI half-widths ±0.086 to ±0.170. Two-judge inter-rater on N=89 corrected cut: Opus 2.72, Sonnet 2.38, per-record Pearson r = 0.814, mean shift -0.39. Bidirectional calibration bracket with Gemma +0.42. ATT&CK technique-to-ID confabulation: 35/36 = 97.2% on name-paired citations, validated against the MITRE Enterprise matrix. Detector TPR collapse on the ttp_summary transform: 20% under Opus and Gemma 4 E4B, 40% under qwen2.5:7b, vs 60-100% on the other two transforms. Opus 4.7 single-record content-specific refusal: 1/30 on N=90, 1/67 on N=194, same chunk_idx=2 ttp_summary record both times. The record falsely attributes attacker-controlled distribution channels to real US-CERT / CISA infrastructure; seven other judges (Sonnet 4.6, Haiku 4.5, Gemma 4 E4B, foundation-sec, qwen2.5:7b, gpt-oss-120b, Llama-4-Scout) all engage. Full analysis in the companion safety-note v4. The trained QLoRA adapter is NOT included in this deposit. The substance-vs-form gap means the adapter is a competent attacker-prose generator with low marginal scientific value beyond what the eval JSONLs already demonstrate; we judge the dual-use cost-benefit unfavourable for open release. The training recipe in scripts/ plus the public-CTI corpus in training_corpus/ are sufficient to reproduce the adapter on a single 16 GB GPU in ~5.5 hours. Reuse: CC-BY-4.0 over the entire deposit. See README.md inside the zip for the bundle layout, file-by-file reproduction recipe, and a diagram of how the harness fits together. Cite the deposit DOI together with the arxiv ID of the preprint when reusing any artefact. Companion deposits: Paper 1 (LLM-targeted recon honeypot measurement) and the safety-classifier note (Opus-specific single-record content guardrail) each get their own Zenodo DOIs; cross-links added on publication.本数据集存档随预印本《无实质的风格：CTI（Cyber Threat Intelligence，网络威胁情报）风格散文的小模型SFT（Supervised Fine-Tuning，监督微调）蒸馏及其对检测器与安全分类器校准的启示》（论文2）一同发布。本存档包含论文源文件、评估工具套件、六维度攻击者角色评分准则，以及三次Opus模型评估结果：原始数据集规模为N=90，重新标注了15条丢失记录，新增规模为N=194的扩展评估数据集；另有针对N=90数据集的Sonnet 4.6跨评分者一致性标注结果、针对N=15数据集的Gemma 4 E4B跨评分者一致性标注结果、ATT&CK框架验证输出、三种分类器下的检测器基线结果，以及包含4010个文本块的公开CTI训练语料库。 v2版本修正了v1版本的方法学错误。v1版本的数据集存档（DOI：10.5281/zenodo.20365823）曾报告，在ttp_summary变换类别下，Opus模型的拒绝率为53%。2026年5月24日的重新调查显示，原报告的16次模型拒绝中，有15次为上游BadRequestError：信用余额不足错误，该类请求在送达Opus模型前就已返回错误；仅1/30为真实的模型拒绝（对应chunk_idx=2的ttp_summary记录，在新增的N=194数据集上复现率为1/67，两次均为同一条记录）。v2版本包含修正后的评估数据、规模为N=194的新增扩展评估数据集（该数据集将各维度的置信区间半宽从±0.13至0.29收紧至±0.086至0.170），以及配套安全说明的v3至v4版本修订轨迹。 本v2数据集存档可复现的核心量化结果如下： 针对N=194扩展评估数据集的Opus 4.7各维度风格保真度均值：整体得分为2.76/5，其中style_consistency（风格一致性）为3.46（唯一达到可接受水平的维度），最低的tradecraft_vocabulary（战术术语词汇量）为2.33。置信区间半宽范围为±0.086至±0.170。 针对N=89修正后数据集的双评分者跨评分者一致性结果：Opus模型得分为2.72，Sonnet模型得分为2.38，逐记录皮尔逊相关系数为0.814，平均分值偏移为-0.39。与Gemma模型的双向校准区间偏移为+0.42。 ATT&CK技术到ID的虚构错误率：在匹配名称的引用中为35/36（即97.2%），已通过MITRE企业矩阵验证。 ttp_summary变换类别下的检测器真阳性率（TPR, True Positive Rate）崩塌现象：在Opus与Gemma 4 E4B模型下为20%，在qwen2.5:7b模型下为40%，而另外两种变换类别下的TPR为60%至100%。 Opus 4.7的单记录内容相关拒绝情况：在N=90数据集下为1/30，在N=194数据集下为1/67，两次均为chunk_idx=2的ttp_summary记录。该记录错误地将攻击者控制的分发渠道归因于真实的美国计算机应急响应小组/网络安全和基础设施安全局（US-CERT / CISA）基础设施；其余7名评分者（Sonnet 4.6、Haiku 4.5、Gemma 4 E4B、foundation-sec、qwen2.5:7b、gpt-oss-120b、Llama-4-Scout）均正常生成了对应内容。完整分析详见配套安全说明v4版本。 本数据集存档未包含训练得到的QLoRA（Quantized Low-Rank Adaptation，量化低秩适配）适配器。由于实质与形式的分离特性，该适配器作为具备良好生成能力的攻击者散文生成器，其边际科学价值已被现有评估JSONL文件覆盖；我们认为其双重使用的成本收益比不利于开源发布。scripts/目录下的训练配方与training_corpus/目录下的公开CTI训练语料库，足以在单张16GB GPU上用约5.5小时复现该适配器。 使用许可：本数据集存档整体采用CC-BY-4.0许可。压缩包内的README.md文件包含数据集打包结构、逐文件复现步骤，以及评估工具套件的工作原理示意图。若使用本数据集的任何组件，请同时引用该数据集的DOI与预印本的arXiv编号。 配套数据集：论文1（针对大语言模型的侦察蜜罐测量）与安全分类器说明（Opus模型专属单记录内容防护栏）均拥有各自的Zenodo DOI；出版后将添加交叉链接。