arXiv subjects

Matt Kretchmar

Publications and source records attributed to Matt Kretchmar.

6 recordsLinked to original sources

arXiv2026

Slow Decay and Silenced Expression: Iterated Subliminal Trait Transfer in Language-Model Lineages

Language models are increasingly trained on the outputs of other models, forming chains that we call lineages, in which a trait present in one generation can pass to the next. Prior work on subliminal learning has shown that a teacher's trait can transmit to a student through filtered data carrying none of the trait's content. However, the evidence covers only a single training step. We study whether such a trait holds or fades across lineages. We instill the trait into three copies of Qwen2.5-7B-Instruct and iterate the training step to depth ten from each, reading every generation two ways on the same held-out prompts: a keyword screen that looks for expressions of the trait in the model's output, and an activation probe that projects each model's displacement from the base onto a direction built from the other lineages' teachers. We report two findings. First, the trait persists through ten generations across three lineages. The instilled models express it on every completion; the keyword-screen rate falls to 55.6% after the first step and to 21.1% by generation ten. The base itself matches the screen on none of its 300 completions. Second, the trait can be present internally while absent behaviorally. When the model's default system prompt is removed at evaluation, the generation-ten students' keyword-screen rate is zero on every prompt while the probe score stays positive on every prompt. Steering the untreated base with the displacement of a generation-ten student, which is trained and measured under the default system prompt, induces screened expression of the trait even with the system prompt removed, while that same student shows no expression of the trait with the system prompt removed.

cs.LG↗

arXiv2024

Towards Robust Extractive Question Answering Models: Rethinking the Training Methodology

This paper proposes a novel training method to improve the robustness of Extractive Question Answering (EQA) models. Previous research has shown that existing models, when trained on EQA datasets that include unanswerable questions, demonstrate a significant lack of robustness against distribution shifts and adversarial attacks. Despite this, the inclusion of unanswerable questions in EQA training datasets is essential for ensuring real-world reliability. Our proposed training method includes a novel loss function for the EQA problem and challenges an implicit assumption present in numerous EQA datasets. Models trained with our method maintain in-domain performance while achieving a notable improvement on out-of-domain datasets. This results in an overall F1 score improvement of 5.7 across all testing sets. Furthermore, our models exhibit significantly enhanced robustness against two types of adversarial attacks, with a performance decrease of only about a third compared to the default models.

cs.CL↗

arXiv2023

AGent: A Novel Pipeline for Automatically Creating Unanswerable Questions

The development of large high-quality datasets and high-performing models have led to significant advancements in the domain of Extractive Question Answering (EQA). This progress has sparked considerable interest in exploring unanswerable questions within the EQA domain. Training EQA models with unanswerable questions helps them avoid extracting misleading or incorrect answers for queries that lack valid responses. However, manually annotating unanswerable questions is labor-intensive. To address this, we propose AGent, a novel pipeline that automatically creates new unanswerable questions by re-matching a question with a context that lacks the necessary information for a correct answer. In this paper, we demonstrate the usefulness of this AGent pipeline by creating two sets of unanswerable questions from answerable questions in SQuAD and HotpotQA. These created question sets exhibit low error rates. Additionally, models fine-tuned on these questions show comparable performance with those fine-tuned on the SQuAD 2.0 dataset on multiple EQA benchmarks.

cs.CL↗

arXiv2023

Single-Sentence Reader: A Novel Approach for Addressing Answer Position Bias

Machine Reading Comprehension (MRC) models tend to take advantage of spurious correlations (also known as dataset bias or annotation artifacts in the research community). Consequently, these models may perform the MRC task without fully comprehending the given context and question, which is undesirable since it may result in low robustness against distribution shift. The main focus of this paper is answer-position bias, where a significant percentage of training questions have answers located solely in the first sentence of the context. We propose a Single-Sentence Reader as a new approach for addressing answer position bias in MRC. Remarkably, in our experiments with six different models, our proposed Single-Sentence Readers trained on biased dataset achieve results that nearly match those of models trained on normal dataset, proving their effectiveness in addressing the answer position bias. Our study also discusses several challenges our Single-Sentence Readers encounter and proposes a potential solution.

cs.CL↗

arXiv2023

The Impacts of Unanswerable Questions on the Robustness of Machine Reading Comprehension Models

Pretrained language models have achieved super-human performances on many Machine Reading Comprehension (MRC) benchmarks. Nevertheless, their relative inability to defend against adversarial attacks has spurred skepticism about their natural language understanding. In this paper, we ask whether training with unanswerable questions in SQuAD 2.0 can help improve the robustness of MRC models against adversarial attacks. To explore that question, we fine-tune three state-of-the-art language models on either SQuAD 1.1 or SQuAD 2.0 and then evaluate their robustness under adversarial attacks. Our experiments reveal that current models fine-tuned on SQuAD 2.0 do not initially appear to be any more robust than ones fine-tuned on SQuAD 1.1, yet they reveal a measure of hidden robustness that can be leveraged to realize actual performance gains. Furthermore, we find that the robustness of models fine-tuned on SQuAD 2.0 extends to additional out-of-domain datasets. Finally, we introduce a new adversarial attack to reveal artifacts of SQuAD 2.0 that current MRC models are learning.

cs.AI↗

arXiv2018

Curriculum Guidelines for Undergraduate Programs in Data Science

The Park City Math Institute (PCMI) 2016 Summer Undergraduate Faculty Program met for the purpose of composing guidelines for undergraduate programs in Data Science. The group consisted of 25 undergraduate faculty from a variety of institutions in the U.S., primarily from the disciplines of mathematics, statistics and computer science. These guidelines are meant to provide some structure for institutions planning for or revising a major in Data Science.

stat.OT↗