arXiv subjects

Allen Chen

Publications and source records attributed to Allen Chen.

5 recordsLinked to original sources

arXiv2026

Improving JEV Entailment Reranking with Span Readouts and Selective Adaptation

JEV decision models pair a language model backbone with a lightweight scoring head to select among natural-language options. Direct selection compares candidate answers within one request, whereas entailment reranking evaluates each candidate in a separate natural language inference (NLI) request over fixed relation options. We examine whether the readout and adaptation choices used for direct selection remain effective for cross-request reranking. We combine a mean-span readout with selective backbone adaptation. Mean-span pooling replaces the option-ending pointer with an average of the contextualized states inside each NLI option, without adding head parameters. Selective adaptation retains low-rank updates in attention and gated DeltaNet projections while removing feed-forward adapters. On MC11, which averages accuracy over eleven reranking configurations from seven sources, selective adaptation improves NeoHorse-1-4B accuracy over the all-module baseline while cutting trainable parameters by more than half, and mean-span pooling yields further consistent gains at the same parameter count.

cs.CL↗

arXiv2026

API-TAIL: Detecting Model Substitution in LLM APIs through Temperature-Aligned Likelihood Tails

Commercial language model APIs return only text, allowing a provider to cut costs by quietly routing a fraction of user requests to a cheaper substitute model while billing for the advertised one. Detecting such partial substitution is challenging because honest providers also alter the output distribution through undocumented decoding and serving choices, such as temperature scaling, sampling truncation, repetition and presence penalties, and default system prompts, forcing an audit to test against a composite null of legitimate configurations rather than a single reference. Our key insight is that these legitimate variations primarily shift response typicality across the entire batch, whereas partial substitution injects individual responses that remain anomalous under the claimed model. Building on this, we introduce API-TAIL, a typicality audit that first absorbs batch-level shifts by aligning a single effective temperature to center standardized surprisal scores across the audited batch, and then flags the endpoint if the number of remaining tail responses exceeds a threshold calibrated across legitimate configurations. Evaluated on real-world API responses and constructed substitution mixtures, API-TAIL controls false alarms across diverse honest providers and detects low-rate partial substitution with a small query budget. Under a common composite calibration, it outperforms the adapted rank- and MMD-based audit baselines.

cs.LG↗

arXiv2025

Malleable Overview-Detail Interfaces

The overview-detail design pattern, characterized by an overview of multiple items and a detailed view of a selected item, is ubiquitously implemented across software interfaces. Designers often try to account for all users, but ultimately these interfaces settle on a single form. For instance, an overview map may display hotel prices but omit other user-desired attributes. This research instead explores the malleable overview-detail interface, one that end-users can customize to address individual needs. Our content analysis of overview-detail interfaces uncovered three dimensions of variation: content, composition, and layout, enabling us to develop customization techniques along these dimensions. For content, we developed Fluid Attributes, a set of techniques enabling users to show and hide attributes between views and leverage AI to manipulate, reformat, and generate new attributes. For composition and layout, we provided solutions to compose multiple overviews and detail views and transform between various overview and overview-detail layouts. A user study on our techniques implemented in two design probes revealed that participants produced diverse customizations and unique usage patterns, highlighting the need and broad applicability for malleable overview-detail interfaces.

cs.HC↗

arXiv2024

Improving QA Model Performance with Cartographic Inoculation

QA models are faced with complex and open-ended contextual reasoning problems, but can often learn well-performing solution heuristics by exploiting dataset-specific patterns in their training data. These patterns, or "dataset artifacts", reduce the model's ability to generalize to real-world QA problems. Utilizing an ElectraSmallDiscriminator model trained for QA, we analyze the impacts and incidence of dataset artifacts using an adversarial challenge set designed to confuse models reliant on artifacts for prediction. Extending existing work on methods for mitigating artifact impacts, we propose cartographic inoculation, a novel method that fine-tunes models on an optimized subset of the challenge data to reduce model reliance on dataset artifacts. We show that by selectively fine-tuning a model on ambiguous adversarial examples from a challenge set, significant performance improvements can be made on the full challenge dataset with minimal loss of model generalizability to other challenging environments and QA datasets.

cs.CL↗

arXiv2024

High-precision Voice Search Query Correction via Retrievable Speech-text Embedings

Automatic speech recognition (ASR) systems can suffer from poor recall for various reasons, such as noisy audio, lack of sufficient training data, etc. Previous work has shown that recall can be improved by retrieving rewrite candidates from a large database of likely, contextually-relevant alternatives to the hypothesis text using nearest-neighbors search over embeddings of the ASR hypothesis text to correct and candidate corrections. However, ASR-hypothesis-based retrieval can yield poor precision if the textual hypotheses are too phonetically dissimilar to the transcript truth. In this paper, we eliminate the hypothesis-audio mismatch problem by querying the correction database directly using embeddings derived from the utterance audio; the embeddings of the utterance audio and candidate corrections are produced by multimodal speech-text embedding networks trained to place the embedding of the audio of an utterance and the embedding of its corresponding textual transcript close together. After locating an appropriate correction candidate using nearest-neighbor search, we score the candidate with its speech-text embedding distance before adding the candidate to the original n-best list. We show a relative word error rate (WER) reduction of 6% on utterances whose transcripts appear in the candidate set, without increasing WER on general utterances.

cs.CL↗