Learn by Surprise, Commit by Proof

• URL: http://arxiv.org/abs/2604.01951v1
• Date: Thu, 02 Apr 2026 12:17:10 GMT
• Title: Learn by Surprise, Commit by Proof
• Abstract summary: We propose a self-gated post-training framework for autonomous knowledge acquisition.<n>When a passage produces anomalously high pertoken loss, L SCP flags it, generates a Q&A chain that forces the model to articulate its own knowledge.<n>The entire learning intensity is governed by a single parameter $r$.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We propose LSCP, a self-gated post-training framework for autonomous knowledge acquisition: learning only what a model does not already know, verified against what it does know, at a strength proportional to conviction, with no external oracle. When a passage produces anomalously high per-token loss, LSCP flags it, generates a Q&A chain that forces the model to articulate its own knowledge and identify gaps, then adjusts AdamW's $β_2$ proportionally to conviction depth k (the number of self-verification steps the passage survives) via $β_2 = 0.999 \cdot r^k$. The entire learning intensity is governed by a single parameter $r$. Beyond new knowledge, this process sharpens weakly encoded existing knowledge, which is a primary source of hallucination. The framework is self-extinguishing: as the model learns, per-token loss on learned passages decreases toward the surprisal threshold and the system progressively converges to standard AdamW. This models biological memory consolidation: temporary information in the context window is selectively consolidated into parametric weights, the model's long-term memory. Experiments on the reference model (Qwen3-14B) and across six models (8B--32B, four families) show that standard fine-tuning produces rote memorization (perturbation gap (the ratio of paraphrase to original perplexity) of 11.6 +- 0.2 x baseline) while all LSCP conditions learn semantically (2.7--3.0x). The r=1.0 condition (identical optimizer, nearly identical data, only Q&A format differs) confirms that the training data format, not $β_2$ gating, is the primary mechanism preventing memorization; gating instead protects neighboring knowledge from contamination by corrupt content (93 +- 7% accuracy on adjacent questions at r=0.98 vs. 90% baseline).

Related papers

• Learning What to Remember: A Cognitively Grounded Multi-Factor Value Model for Agentic Memory [1.2127875744950842]
  Long-running LLM agents accumulate interaction histories far larger than any context window.<n>Production systems answer with semantic similarity or recency.<n>We propose a multi-factor memory function V(m)=sum_i w_i f_i(m) over seven interpretable factors.
  arXiv  Detail & Related papers  (2026-06-11T06:17:31Z)
• Lying Is Just a Phase: The Hidden Alignment Transition in Language Model Scaling [0.0]
  We measure the coupling between reasoning and truthfulness across 63 base models from 16 families.<n>We find a regime change invisible to loss curves below a family-dependent critical scale $N_c$, capabilities anticorrelate; above it, they cooperate.
  arXiv  Detail & Related papers  (2026-05-13T03:14:09Z)
• The Geometry of Forgetting: Temporal Knowledge Drift as an Independent Axis in LLM Representations [50.43168858368539]
  Large language models confidently produce outdated answers, and no existing method can detect them.<n>We show this is not an engineering failure but a structural one: temporal drift is encoded as a direction in the residual stream geometrically to both correctness and uncertainty.
  arXiv  Detail & Related papers  (2026-05-09T22:27:31Z)
• Learning the Signature of Memorization in Autoregressive Language Models [3.6048665052465663]
  We introduce the first transferable learned attack, enabled by the observation that fine-tuning any model on any corpus yields unlimited labeled data.<n>This removes the shadow model bottleneck and brings membership inference into the deep learning era.
  arXiv  Detail & Related papers  (2026-04-03T17:17:51Z)
• Self-Calibrating Language Models via Test-Time Discriminative Distillation [18.46710400838861]
  Large language models (LLMs) are systematically overconfident on questions they often answer incorrectly.<n>We introduce $textbfSECL$ ($textbfSE$lf-$textbfC$alibrating $textbfL$anguage Models), a test-time training (TTT) pipeline that exploits this gap as label-free self-supervision.
  arXiv  Detail & Related papers  (2026-03-18T13:28:50Z)
• Probing for Knowledge Attribution in Large Language Models [45.47366023067617]
  Large language models (LLMs) often generate fluent but unfounded claims, or hallucinations.<n> Proper mitigation depends on knowing whether a model's answer is based on the prompt or its internal weights.<n>We show that a probe, a simple linear classifier trained on model hidden representations, can reliably predict contributive attribution.
  arXiv  Detail & Related papers  (2026-02-26T09:21:12Z)
• Representation-Aware Unlearning via Activation Signatures: From Suppression to Knowledge-Signature Erasure [2.0017902634527194]
  We introduce Knowledge Immunization Framework (KIF), a representation-aware architecture that distinguishes genuine erasure from obfuscation.<n>Our approach combines dynamic suppression of subject-specific representations with parameter-efficient adaptation, enabling durable unlearning without full model retraining.
  arXiv  Detail & Related papers  (2026-01-15T16:28:14Z)
• Assimilation Matters: Model-level Backdoor Detection in Vision-Language Pretrained Models [71.44858461725893]
  Given a model fine-tuned by an untrusted third party, determining whether the model has been injected with a backdoor is a critical and challenging problem.<n>Existing detection methods usually rely on prior knowledge of training dataset, backdoor triggers and targets.<n>We introduce Assimilation Matters in DETection (AMDET), a novel model-level detection framework that operates without any such prior knowledge.
  arXiv  Detail & Related papers  (2025-11-29T06:20:00Z)
• LLMs Can Get "Brain Rot"! [68.08198331505695]
  Continual exposure to junk web text induces lasting cognitive decline in large language models (LLMs)<n>We run controlled experiments on real Twitter/X corpora, constructing junk and reversely controlled datasets.<n>Results provide significant, multi-perspective evidence that data quality is a causal driver of LLM capability decay.
  arXiv  Detail & Related papers  (2025-10-15T13:28:49Z)
• Valid Stopping for LLM Generation via Empirical Dynamic Formal Lift [6.908972852063454]
  Sequential-EDFL reduces generation by 22-28% vs. sequential baselines.<n>EDFL serves as a first-stage filter reducing verification burden by 83%.
  arXiv  Detail & Related papers  (2025-10-07T21:28:53Z)
• R-Tuning: Instructing Large Language Models to Say `I Don't Know' [66.11375475253007]
  Large language models (LLMs) have revolutionized numerous domains with their impressive performance but still face their challenges. Previous instruction tuning methods force the model to complete a sentence no matter whether the model knows the knowledge or not. We present a new approach called Refusal-Aware Instruction Tuning (R-Tuning) Experimental results demonstrate R-Tuning effectively improves a model's ability to answer known questions and refrain from answering unknown questions.
  arXiv  Detail & Related papers  (2023-11-16T08:45:44Z)
• Physics of Language Models: Part 3.1, Knowledge Storage and Extraction [51.68385617116854]
  Large language models (LLMs) can store a vast amount of world knowledge, often extractable via question-answering. We find a strong correlation between the model's ability to extract knowledge and various diversity measures of the training data.
  arXiv  Detail & Related papers  (2023-09-25T17:37:20Z)
• Information-Theoretic Bayes Risk Lower Bounds for Realizable Models [11.203656874463697]
  We derive information-theoretic lower bounds on the Bayes risk and generalization error of realizable machine learning models. For realizable models, we show that both the rate distortion function and mutual information admit expressions that are convenient for analysis.
  arXiv  Detail & Related papers  (2021-11-08T15:42:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.