ActMap is introduced, a white-box representation that compresses the generation-time hidden-state trajectory into a fixed tensor of temporal-statistic channels that preserves structure across transformer depth and pooled hidden coordinates, and supports abstention, routing, and selective verification from a single generation, a practical primitive for scalable oversight of deployed models.
Abstract
Practical uncertainty quantification (UQ) for large language models must decide, from a single generation, whether a specific answer should be trusted. Existing methods either sample multiple generations, read only output-token probabilities, or reduce the model's internal computation to a single hidden state. We introduce ActMap, a white-box representation that compresses the generation-time hidden-state trajectory (every layer, every generated token) into a fixed $12\times32\times128$ tensor of temporal-statistic channels that preserves structure across transformer depth and pooled hidden coordinates. The map is captured during the generation pass with no measurable overhead, has a fixed shape across model depths and hidden sizes, and occupies 96 KiB: a compact artifact that can be retained for audit-relevant generations and probed directly, with occlusion analysis localizing the classifier's signal to mid-depth regions of the map. A lightweight classifier, instantiated as a compact Vision Transformer, reads an estimated correctness probability from each map in a fraction of a millisecond; capacity-matched MLPs perform comparably, indicating the representation itself carries the result. Trained and evaluated in-domain on short-answer QA, direct-answer math, and summarization factuality with three instruction-tuned 7-8B models, ActMap consistently outperforms sampling, token-probability, attention, and embedding baselines, and matches ACT-ViT, a detector trained on dense activation tensors $67\times$ larger, at essentially the same mean AUROC with lower calibration error on ten of twelve pairs. The resulting score supports abstention, routing, and selective verification from a single generation, making it a practical primitive for scalable oversight of deployed models.
HSRM is introduced, a lightweight hidden-state reward model that verifies candidate solutions by directly reading the generator's internal representations rather than re-processing its text, providing an efficient alternative to text-only verification by reusing representations already computed during generation.
This work test whether action maps fitted without a source reach its natural post-action activation and compose and validate the geometric branch on a known affine S_5 carrier: all held-source folds pass one-step, composition, inverse, decoding, and commutativity gates.
End-to-end activation-state transfer between LLMs, as currently implemented, is architecture-dependent rather than universal, and it is concluded that end-to-end activation-state transfer between LLMs is architecture-dependent rather than universal.
Low-bit vision-language-action inference must reduce observation-to-action latency while preserving robot behavior. We present FoldQuantVLA, a post-training quantization framework that carries a consistent activation representation through calibration, weight rounding, and native integer execution. It combines channel...
Hung T. Ho, Khanh-Binh Nguyen, Q. Nguyen et al.· arXiv.org· 1 citation
It is suggested that reasoning validity is better read from state-conditioned motion than from either static states or decontextualized trajectories alone, andlations show that motion, region, and direction provide complementary signals.
Hamed Damirchi, I. M. De La Jara, D. Ranasinghe et al.· 0 citations
It is found that only customizing the threshold for exiting early is worthwhile: calibrating it per deployment raised exit rates by up to 59 percentage points across three models, and by more than 10 points on most corpora tested, while other techniques showed far less impact.