Internal-state probes enable truthfulness prediction before a large language model generates an answer. When detectors change both the layers they read and the rules used to combine them, the source of improved prediction becomes difficult to identify. We separate these choices and find that retaining more layers impro...
Zhi-Hao Guo, Zong-Han Wu, Huan Huo et al.· 0 citations
Across 13 prediction tasks spanning post-transcriptional regulation, RNA fate and expression output, EukaUTR models matched or exceeded the strongest external baselines on nearly all tasks, with relative improvements of up to 27.45%.
Mei Lang, Xing-Yu Fang, Ming-Xuan Chen et al.· bioRxiv· 0 citations
CodonMamba is presented, a codon language model framework for mRNA prediction and programmable CDS design, enabling context-specific mRNA sequence engineering and highlighting a promising direction for precise mRNA design using foundation models.
Mei Lang, Xing-Yu Fang, Zhen Wang et al.· bioRxiv· 0 citations
HalluTracer is introduced, a detection framework that reads and aggregates truthfulness evidence across every layer of the forward pass before the model emits any answer token, recasts hallucination detection from a layer-selection problem into a depth-aggregation problem governed by the geometric sparsity of the truth...
Zhi-Hao Guo, Zong-Han Wu, Huan Huo et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
