Jul 2026· Anais do LIII Seminário Integrado de Software e Hardware (SEMISH 2026)· pp. 914-919· 0 citations· 6 references
TL;DR
ParameterEfficient Fine-Tuning (PEFT) mitigates data leakage in Llama and Gemma and Qwen remain highly vulnerable across all adaptation methods (p < 0.05), while Gemma and Qwen remain highly vulnerable across all adaptation methods.
Abstract
Dataset Inference provides a robust framework for auditing data ownership by aggregating statistical signals that traditional Membership Inference Attacks (MIAs) fail to capture. While proven for LLM pre-training, its efficacy during fine-tuning is largely unexplored. We evaluate Dataset Inference on Gemma, Llama, and Qwen models using Full Fine-Tuning (FFT), LoRA, and QLoRA. Our findings reveal a stark architectural divergence: ParameterEfficient Fine-Tuning (PEFT) mitigates data leakage in Llama (AUC ≈ 0.50, p > 0.05), while Gemma and Qwen remain highly vulnerable across all adaptation methods (p < 0.05). Additionally, higher learning rates accelerate data absorption, and QLoRA provides only marginal regularization compared to standard LoRA.
Diffusion language models (DLMs) have recently emerged as an alternative modeling paradigm to autoregressive LMs, offering advantages such as parallel generation and bidirectional context modeling. Despite growing interest in their generative capabilities, the privacy risks of DLMs remain underexplored. We identify a phenomenon termed token-level memorization asymmetry through theoretical analysis of diffusion training dynamics. Building on this finding, we propose Q-Skew, a quantile-weighted skewness-based indicator for membership inference on finetuned DLMs. Experiments across multiple fine-tuning datasets and models show that our method outperforms existing baselines. Moreover, we show that Q-Skew can also facilitate other privacy violations, such as PII extraction. Our findings reveal a previously underexplored privacy attack surface and highlight the need for systematic privacy evaluation of DLMs.
Shengfang Zhai, Leo Marchyok, Yu-Ling Shi et al.· 0 citations
We evaluate three open-weight LLMs (Gemma3-12B from the USA, Bielik-11B-v3 from Poland, and Qwen3-4B from China) against World Values Survey Wave 7 data for 63 demographic personas across three countries, using normalized Wasserstein distance to quantify distributional misalignment. Contrary to expectations, no model favors its home country: the Chinese-built Qwen3-4B performs worst on its own Chinese population (W1 = 0.436, the highest misalignment in the entire model x country matrix). Targeted LoRA fine-tuning on the five worst-case personas, requiring fewer than 1,200 training pairs and under 15 minutes on a single GPU, reduces bias by 16.8% for Bielik-11B (p_Bonf = 0.002, d = -4.4) with all five targets improving. However, country-level decomposition reveals that fine-tuning redistributes rather than removes bias: Bielik's worst-case personas swap entirely from American to Chinese elderly, with zero overlap between pre- and post-correction sets. To our knowledge, this is the first study to target worst-case demographic personas with LoRA fine-tuning for cross-cultural bias mitigation.
It is argued that WG is more plausible as an adversarial threat-requiring careful data engineering-rather than as a significant hazard inherent to routine fine-tuning.
Benchmark scores are reported as properties of a model, yet the inference framework used to produce them, such as HuggingFace, vLLM, or Ollama, are considered non-influential and their names and versions are almost never disclosed. In this work we investigate how much this choice can influence the model output. In a fully-crossed study (three instruction-tuned models x five inference frameworks x six benchmarks x four generation modes) we investigate how different tools (wrappers/backend) influence benchmark scores and how their score changes is influenced by generation hyper-parameters. We find backend to be a non-negligible factor where even under greedy, sampling-noise-free decoding, changing the backend can significantly alter models performance and this effect is structural and strongly model-dependent. Decomposing the variance according to generation mode reveal that considerable portion of the variability (roughly 39\%) a practitioner sees out-of-the-box can stem from the backend, while the remaining stems from sampling noise and each framework's default generation parameters, both of which are avoidable by disclosing and matching the generation configuration. These divergences are more pronounced on factual than on social-bias benchmarks. Overall, benchmark numbers are not backend-agnostic therefore, we recommend disclosing the backend, its version, and the full generation configuration, also using deterministic decoding for cross-backend comparison.
Shahed Masoudian, Passant Shafaei, Monorama Swain et al.· 0 citations
Membership inference attacks (MIAs) try to determine whether a specific record was used to train a model, a privacy risk that matters in natural language processing (NLP), where training data can contain sensitive user text. This paper presents a controlled benchmark of membership inference vulnerability for text classification on the GLUE SST-2 sentiment dataset. A TF-IDF + Logistic Regression pipeline and a fine-tuned DistilBERT classifier are compared under a loss-threshold MIA, with utility measured by development accuracy and macro F1. DistilBERT reached 0.9466 accuracy and 0.9460 macro F1 against 0.8756 and 0.8727 for Logistic Regression, yet both models leaked membership signal (Attack AUC 0.5615 and 0.5800, respectively). Two mitigations were tested. Stronger regularization reduced leakage for Logistic Regression at a visible utility cost, whereas fine-tuning DistilBERT for 2 epochs instead of 3 reduced leakage with negligible accuracy loss. Lightweight training adjustments can improve the privacy-utility trade-off without complex defenses.
William Novak (Minot State University), Muhammad Abusaqer (Minot State University)· 0 citations
TopoTuner is competitive with full fine-tuning while training only 1-2% of the model parameters, and outperforms LoRA in 7 out of 9 model-dataset settings, which can change up to 39.57% of the projection parameters.
