EdTech Discovery
Argus

Named after the hundred-eyed watchman of Greek myth, Argus watches the education landscape: spotting new opportunities, pressure-testing the ventures we're building, and tracing every read back to the real-world signals behind it.

Updated Sep 07, 2026 · 40 ideas · 18624 signals
Admin mode. Curation controls visible. Keep this URL (with token) private.

Signals

The evidence library: the raw signals the pipeline is watching across the education ecosystem. Every idea is built from these.

technology Mon, 20 Jul 2026 21:25:27 +0000
MedCity News

Expecting Massive Test Growth in Cancer & More, Tempus AI Strikes $1.5B Deal to Buy Personalis

Personalis’s liquid biopsy is part of a growing market for minimal residual disease assays, tests that look for any remaining cancer cells after the initial treatment. Tempus AI has been the commercial partner on this Personalis test since 2023. The post Expecting Massive Test Growth in Cancer & More, Tempus AI Strikes $1.5B Deal to Buy Personalis appeared first on MedCity News .

Source ↗
technology Mon, 20 Jul 2026 16:37:16 +0000
HN: education

Why Did We Let Big Tech Take over Education?

Article URL: https://www.afterbabel.com/p/sophie-winkleman-edtech-rollback-2026 Comments URL: https://news.ycombinator.com/item?id=48981217 Points: 8 # Comments: 0

Source ↗
technology Mon, 20 Jul 2026 16:22:37 -0400
EdTech Mag (Higher)

How Florida State University Is Building Its AI Ecosystem

Faculty and students, academic researchers, the staff and administrators working behind the scenes at colleges and universities around the country are all leveraging artificial intelligence to teach, learn, and work more efficiently and effectively. One recent survey, in fact, found that 95% of students and educators are actively using AI tools. And while the surge in AI adoption in higher education kicked off in 2022 with OpenAI’s release of ChatGPT, today it’s being driven by a wide range of platforms from established and emerging companies alike. In this environment, a growing number of…

Source ↗
technology Mon, 20 Jul 2026 14:04:27 -0400
EdTech Mag (K-12)

School’s Out: Focus on Device Maintenance and Readiness for Fall

For K–12 IT teams, summer is one of the busiest and most important times of the year. Once devices are collected, classrooms are cleared and summer projects begin, technology teams must assess, repair, update and prepare the tools that students and teachers will rely on when they return to school. Classroom technology use continues to grow, driven by new tools. Chromebooks and other personal devices are transported between home and school, and not always in the most careful ways. By the time June arrives, devices show signs of use. There may be missing keys, cracked or damaged screens,…

Source ↗
technology Mon, 20 Jul 2026 13:43:00 +0000
MedCity News

The AI Tension in Healthcare: Patent Strategy, FDA Reality, and HIPAA Constraints

These three legal regimes are pulling in different directions, and the developers that navigate them well will be the ones that plan for all three from the start. The post The AI Tension in Healthcare: Patent Strategy, FDA Reality, and HIPAA Constraints appeared first on MedCity News .

Source ↗
technology Mon, 20 Jul 2026 13:38:51 +0000
MedCity News

For Millions in Rural America, Medicine Is Still Far Away

Rural Health Transformation Program funding gives states broad flexibility and may help address longstanding access gaps. The post For Millions in Rural America, Medicine Is Still Far Away appeared first on MedCity News .

Source ↗
technology Mon, 20 Jul 2026 09:30:00 +0000
Tech & Learning

From Pitch to Partnership: A New Playbook for Marketing to School Districts

Insights education companies can use to evolve their marketing and sales strategies and build lasting partnerships with schools

Source ↗
technology Mon, 20 Jul 2026 09:00:00 +0000
Tech & Learning

What is Aqua from Adobe and How Can I Use It To Teach?

Aqua from Adobe is a free drawing and illustrator tool designed specifically for kids.

Source ↗
technology Mon, 20 Jul 2026 09:00:00 +0000
eCampus News

What “friction maxxing” teaches us about IT skill development in the age of AI

As AI adoption accelerates, a countertrend has emerged that reinforces the old saying: What goes around comes around. After spending years looking for ways to remove friction from everyday life, a growing number of people are embracing it by choosing to perform certain tasks without the assistance of AI and other forms of automation. The post What “friction maxxing” teaches us about IT skill development in the age of AI appeared first on eCampus News .

Source ↗
technology Mon, 20 Jul 2026 02:18:18 +0000
MedCity News

Women’s Oncology Market Could Reach $110B in 2030, but Gaps Remain

A new PwC report says the women’s oncology market could reach $110 billion by 2030, though funding and care gaps remain. The post Women’s Oncology Market Could Reach $110B in 2030, but Gaps Remain appeared first on MedCity News .

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

NAMESAKES: Probing Identity Memorization in Text-to-Image Models

arXiv:2606.20155v2 Announce Type: replace-cross Abstract: Text-to-image (T2I) models generate realistic likenesses of some individuals when prompted with their names, raising privacy concerns. However, distinguishing whether a generated face is memorized or fabricated currently requires ground-truth photos, access to training data, or white-box access to model internals, limiting applicability. We introduce a fully black-box behavioral probe that distinguishes between memorized and unrecognized names, while requiring no reference photos or prior knowledge of training data. To benchmark this task, we present the NAMESAKES dataset of over one thousand names and faces of public figures spanning a wide range of fame levels, along with perturbed, less famous names. Experiments on state-of-the-art T2I models show that our probe substantially predicts identity memorization and separates memorized from unrecognized names, with further insights into differences across model families.

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Crayotter: Traceable Multi-Agent Workflows for Long-Form Video Editing

arXiv:2606.07636v2 Announce Type: replace-cross Abstract: Long-form video editing over heterogeneous footage requires agents to coordinate source selection, multimodal analysis, timeline construction, narration and subtitle alignment, rendering, and revision while exposing intermediate state for inspection and repair. We present Crayotter, an open-source multimodal multi-agent demo system for prompt-driven long-form video editing. Crayotter organizes production around coverage-aware material preparation, artifact-grounded editing research, and tool-grounded timeline execution. Across these stages, retrieval reports, video analyses, editing blueprints, scheduler events, tool calls, intermediate renders, and final exports are treated as first-class artifacts rather than hidden transient state. The workbench supports local assets, agent-assisted retrieval, progress monitoring, artifact preview, failure diagnosis, interrupted-job resumption, and resource-aware asynchronous execution for lo

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

AgentRedBench: Dynamic Redteaming and Integration-Aware Defense for LLM Agents over SaaS Integrations

arXiv:2606.02240v3 Announce Type: replace-cross Abstract: Indirect prompt injection in tool-use agents is a concrete production threat: LLM agents read from integrations (third-party services such as Gmail, Salesforce, or Jira accessed through tool calls) whose response content the user neither writes nor controls. Existing benchmarks under-measure the threat: most cover only a handful of integrations with the same attack payload replayed across runs, and open-source guards are trained on chat-style data rather than tool-response content. We introduce AGENTREDBENCH, a dynamic LLM-driven redteaming benchmark of 215 subtle underspecified-authorization scenarios across 24 enterprise integrations and five attack types. Across an eight-model panel (Anthropic, OpenAI, Google), no-guard attack success rate ranges from 32% to 81%. To keep the scenario set out of training corpora and preserve headline ASR meaning over time, we release the codebase, integration schemas, and AGENTREDGUARD model o

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Self-Attention as Transport: Limits of Symmetric Spectral Diagnostics

arXiv:2605.04893v3 Announce Type: replace-cross Abstract: Every attention head defines a degree-normalized transport operator, and a growing family of diagnostics reads model behavior (hallucination among them) from its spectrum. We ask what such diagnostics can and cannot infer. The operator splits orthogonally into a symmetric part governing transport \emph{capacity} and an antisymmetric part encoding \emph{orientation}. We prove an identifiability limit: every transpose-invariant spectral diagnostic is \emph{orientation-blind} (unable to distinguish an operator from its transpose, hence blind to the orientation of information flow), with a transpose-stability bound limiting any Lipschitz diagnostic's transpose sensitivity by the asymmetry coefficient $G$. This bounds what spectral diagnostics of the attention operator can resolve (e.g.\ LapEigvals and the attention-spectral branch of LLM-Check). On the surviving axis, a closed-form bipartite-Cheeger landscape shows uniform causal at

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

What Is the Minimum Architecture for Prolepsis? Early Irrevocable Commitment Across Tasks in Small Transformers

arXiv:2604.15010v2 Announce Type: replace-cross Abstract: When do transformers commit to a decision, and what prevents them from correcting it? We introduce prolepsis: a transformer commits early, task-specific attention heads sustain the commitment, and no layer corrects it. Replicating Lindsey et al.'s (2025) planning-site finding on open models (Gemma 2 2B, Llama 3.2 1B), we ask five questions. (Q1) Planning is invisible to six residual-stream methods; among those tested, only CLT-based steering succeeds. (Q2) The single-site spike replicates in shape, at the final prompt token (Anthropic's site is the newline; see the Note added). (Q3) Specific attention heads route the decision to the output, filling a gap flagged as invisible to attribution graphs. (Q4) The evidence is consistent with search within at most 16 layers and commitment beyond, a two-model hypothesis. (Q5) Factual recall shows the same motif at a different network depth, with zero overlap between recurring planning hea

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Ruling Out to Rule In: Contrastive Hypothesis Retrieval for Medical Question Answering

arXiv:2604.04593v2 Announce Type: replace-cross Abstract: Retrieval-augmented generation (RAG) grounds large language models in external medical knowledge, yet standard retrievers frequently surface hard negatives that are semantically close to the query but describe clinically distinct conditions. While existing query-expansion methods improve query representation to mitigate ambiguity, they typically focus on enriching target-relevant semantics without an explicit mechanism to selectively suppress specific, clinically plausible hard negatives. This leaves the system prone to retrieving plausible mimics that overshadow the actual diagnosis, particularly when such mimics are dominant within the corpus. We propose Contrastive Hypothesis Retrieval (CHR), a framework inspired by the process of clinical differential diagnosis. CHR generates a target hypothesis $H^+$ for the likely correct answer and a mimic hypothesis $H^-$ for the most plausible incorrect alternative, then scores document

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

SODA: Semi On-Policy Black-Box Distillation for Large Language Models

arXiv:2604.03873v4 Announce Type: replace-cross Abstract: Black-box knowledge distillation for large language models presents a strict trade-off. Simple off-policy methods (e.g., sequence-level knowledge distillation) struggle to correct the student's inherent errors. Fully on-policy methods (e.g., Generative Adversarial Distillation) solve this via adversarial training but introduce well-known training instability and crippling computational overhead. To address this dilemma, we propose SODA (Semi On-policy Distillation with Alignment), a highly efficient alternative motivated by the inherent capability gap between frontier teachers and much smaller base models. Because a compact student model's natural, zero-shot responses are almost strictly inferior to the powerful teacher's targets, we can construct a highly effective contrastive signal simply by pairing the teacher's optimal response with a one-time static snapshot of the student's outputs. This demonstrates that exposing the sma

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Jailbreak Foundry: From Papers to Runnable Attacks for Reproducible Benchmarking

arXiv:2602.24009v4 Announce Type: replace-cross Abstract: Jailbreak techniques for large language models (LLMs) evolve faster than benchmarks, making robustness estimates stale and difficult to compare across papers due to drift in datasets, harnesses, and judging protocols. We introduce JAILBREAK FOUNDRY (JBF), a system that addresses this gap via a multi-agent workflow to translate jailbreak papers into executable modules for immediate evaluation within a unified harness. JBF features three core components: (i) JBF-LIB for shared contracts and reusable utilities; (ii) JBF-FORGE for the multi-agent paper-to-module translation; and (iii) JBF-EVAL for standardizing evaluations. Across 30 reproduced attacks, JBF achieves high fidelity with a mean (reproduced-reported) attack success rate (ASR) deviation of +0.26 percentage points. By leveraging shared infrastructure, JBF reduces attack-specific implementation code by more than half relative to original repositories and achieves an 82.5%

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Mechanistic Interpretability of Cognitive Complexity in LLMs via Linear Probing using Bloom's Taxonomy

arXiv:2602.17229v2 Announce Type: replace-cross Abstract: The black-box nature of Large Language Models necessitates novel evaluation frameworks that transcend surface-level performance metrics. This study investigates the internal neural representations of cognitive complexity using Bloom's Taxonomy as a hierarchical lens. By analyzing high-dimensional activation vectors from different LLMs, we probe whether different cognitive levels, ranging from basic recall (Remember) to abstract synthesis (Create), are linearly separable within the model's residual streams. Our results demonstrate that linear classifiers achieve approximately 95% mean accuracy across all Bloom levels, providing strong evidence that cognitive level is encoded in a linearly accessible subspace of the model's representations. These findings provide evidence that the model resolves the cognitive difficulty of a prompt early in the forward pass, with representations becoming increasingly separable across layers.

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

LLM-as-a-Judge Scores Are Unreliable Optimization Signals in Closed-Loop Table Recognition

arXiv:2607.13347v2 Announce Type: replace Abstract: LLM-as-a-judge is widely used to provide feedback and selection signals in closedloop regeneration, but this use remains insufficiently validated. We study it in table recognition, where deterministic TEDS evaluation provides a controlled testbed, using FinTabNet and OmniDocBench. Three findings emerge. First, judge signals were weak on both datasets: scores frequently tied, rankings were not reproducible, and no tested judge score policy, whether selecting among candidates or accepting revisions under a conservative score margin, improved on the first output on both datasets. Iteration produced better candidates, but the judge recovered them at most partially on one dataset and not at all on the other. Second, severe losses occurred even without specific judge feedback, supporting target-preservation failure under unconstrained regeneration as a proximate mechanism. Third, a structure-preserving instruction reduced the severe-loss ra

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

An expressivity analysis of hierarchical modelling in deep transformers via bounded-depth grammars

arXiv:2606.17522v2 Announce Type: replace Abstract: Deep neural networks are widely believed to derive their expressive power from their ability to form \textbf{hierarchical representations}, capturing progressively more abstract and compositional features across layers. In language modeling, \textbf{transformers} have emerged as the dominant architecture, with early layers capturing local syntactic patterns and later layers encoding more complex clause-level dependencies. While this intuition has shaped model design, there remains a lack of rigorous theoretical work demonstrating \textbf{how} deep transformers represent such hierarchical structures. In this work, we analyze the expressiveness of deep transformer models through the formal lens of bounded-depth, non-recursive context-free grammars. For this class of grammars, we explicitly construct transformers with positional attention whose depth grows linearly with grammar depth, while the neuron count scales with the number of deri

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

AuAu: A Benchmark for Auditing Authoritarian Alignment in Large Language Models

arXiv:2606.16127v2 Announce Type: replace Abstract: The worldwide rise of authoritarianism and the growing role of Large Language Models (LLMs) in users' everyday lives raise the question of whether specific models exhibit or promote authoritarian attitudes. We introduce AuAu, a comprehensive benchmark for assessing the risk of authoritarian tendencies in LLM responses. AuAu combines three evaluation approaches: (i) psychometric questions from 15 human-validated instruments, (ii) vignettes probing intended behavior in concrete situations, and (iii) responses to realistic user prompts. Unlike prior work, AuAu measures not only overall authoritarian alignment but also its established sub-concepts: Authoritarian Aggression, Authoritarian Submission, and Conventionalism. Evaluating 17 models from China, the EU, Russia, and the USA, we find substantial authoritarian response rates on psychometric instruments across all models, though rates drop significantly on more realistic downstream tas

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

It Takes One to Bias Them All: Breaking Bad with One-Shot GRPO

arXiv:2606.10931v3 Announce Type: replace Abstract: Warning: This paper contains several toxic and offensive statements. Modern large language models (LLMs) are typically aligned through large-scale post-training to ensure fair and reliable behavior. In this work, we investigate how easily such guardrails can be broken by Group Relative Policy Optimization (GRPO). We show that one-shot GRPO training on a single biased example is sufficient to induce systematic bias, with stereotype-driven reasoning generalizing across attributes, categories, and benchmarks. We further find that models differ in their susceptibility based on the initial likelihood of producing biased outputs. Our results reveal a critical vulnerability in post-training: alignment can be overridden by a single example.

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

What Should a Skill Remember? Quality--Cost Trade-offs in Cost-Aware Skill Rewriting for Language Model Agents

arXiv:2606.09421v3 Announce Type: replace Abstract: Large language model agents increasingly rely on skills: reusable procedural documents encoding workflows, tool use, implementation patterns, validation checks, and domain rules. Skill rewriting is often treated as prompt compression, but shorter skills can make agents more expensive by removing sparse operational anchors that prevent exploration, debugging, and recovery. We study skill rewriting through this economic lens. Our controlled framework profiles skill structure, rewrites skills using information-preservation strategies, and evaluates the rewrites under fixed task instructions, environments, and verifiers. Experiments on SkillsBench reveal distinct quality--cost trade-offs across strategies: API/code anchoring, workflow guarding, and rule/formula anchoring benefit different task families, with no universally dominant template. In the main held-out evaluation, the learned policy reduces total cost by 7.0% and downstream agen

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Probing LLMs for Syntactic Structure Beyond Universal Dependencies: A Minimalist Phase Account in English

arXiv:2605.26431v3 Announce Type: replace Abstract: We show that LLMs encode syntactic distinctions not present in the Universal Dependencies (UD) tree distances that structural probes are trained to recover. On English wh-movement stimuli, we measure the probe distance between an embedded subject and its verb, whose UD tree distance is invariant across conditions. That distance is shorter than baseline when the embedded clause is finite and longer when it is infinitival -- a within-clause sign asymmetry present in all 13 models across four families we test, at a majority of layers. No account based on UD distance, linear order, or monotone structural complexity can produce a sign reversal, while Minimalist phase theory can. Holding the matrix verb fixed while varying only the complement type reproduces the same finite-infinitival ordering in every model, ruling out a lexical-semantic explanation. The cross-clause pair separately reproduces the phase-count ordering of earlier work, val

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

VCG-Bench: Towards A Unified Visual-Centric Benchmark for Structured Generation and Editing

arXiv:2605.15677v2 Announce Type: replace Abstract: Despite the rapid advancements in Vision-Language Models (VLMs), a critical gap remains in their ability to handle structured, controllable diagrammatic tasks essential for professional workflows. Existing methods predominantly rely on pixel-based synthesis, which operates in probabilistic pixel spaces and is inherently limited in editability and fidelity. Instead, we propose a new Diagram-as-Code paradigm with symbolic logic that leverages mxGraph Extensible Markup Language (XML) for precise diagram generation and editing. We present VCG-Bench, a unified benchmark for visual-centric \texttt{mxGraph} tasks. VCG-Bench comprises: (1) a taxonomized dataset of 1,449 diverse diagrams spanning 6 domains and 15 sub-domains, (2) a paradigm definition that integrates Generation (Vision-to-Code) and Editability (Code-to-Code), (3) a Tailored Evaluation Protocol employing multi-dimensional metrics such as \texttt{mxGraph} Execution Success Rate,

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

From Articles to Premises: Building PrimeFacts, an Extraction Methodology and Resource for Fact-Checking Evidence

arXiv:2605.06006v2 Announce Type: replace Abstract: Fact-checking articles encode rich supporting evidence and reasoning, yet this evidence remains largely inaccessible to automated verification systems due to unstructured presentation. We introduce PrimeFacts, a methodology and resource for extracting fine-grained evidence from full fact-checking articles. We compile 13,106 PolitiFact articles with claims, verdicts, and all referenced sources, and we identify 49,718 in-article hyperlinks as natural anchors to pinpoint key evidence. Our framework leverages large language models (LLMs) to rewrite these anchor sentences into stand-alone, context-independent premises and investigates the extraction of additional implicit evidence. In evaluations on cross-article evidence retrieval and claim verification, the extracted premises substantially improve performance. Decontextualized evidence yields higher retrievability, achieving up to a 30 percent relative gain in Mean Reciprocal Rank over v

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

RLearner-LLM: Balancing Logical Grounding and Fluency in Large Language Models via Hybrid Direct Preference Optimization

arXiv:2605.04539v4 Announce Type: replace Abstract: Direct Preference Optimization (DPO), the efficient alternative to PPO-based RLHF, falls short on knowledge-intensive generation: standard preference signals from human annotators or LLM judges exhibit a systematic verbosity bias that rewards fluency over logical correctness. This blindspot leaves a logical alignment gap -- SFT models reach NLI entailment of only 0.05-0.22 despite producing fluent text. We propose RLearner-LLM with Hybrid-DPO: an automated preference pipeline that fuses a DeBERTa-v3 NLI signal with a verifier LLM score, removing human annotation while overcoming the "alignment tax" of single-signal optimization. Evaluated across five academic domains (Biology, Medicine, Law) with three base architectures (LLaMA-2-13B, Qwen3-8B, Gemma 4 E4B-it), RLearner-LLM yields up to 6x NLI improvement over SFT, with NLI gains in 11 of 15 cells and consistent answer-coverage gains. On Gemma 4 E4B-it (4.5B effective params), Hybrid-

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Brain-CLIPLM: Semantic Compression for EEG-to-Text Decoding

arXiv:2604.16370v3 Announce Type: replace Abstract: Decoding natural language from non-invasive electroencephalography (EEG) remains constrained by low signal-to-noise ratio and limited information bandwidth. This raises a central question: can sentence-level language be reliably recovered from such signals? Under realistic information constraints, this direct-recovery assumption may be too strong. We introduce a semantic compression hypothesis: non-invasive EEG may preserve recoverable semantic anchors rather than the full lexical--syntactic form of a sentence. From this perspective, direct sentence reconstruction is overly fine-grained relative to the recoverable information scale of EEG. To address this mismatch, we propose Brain-CLIPLM, a two-stage framework that decomposes EEG-to-text decoding into semantic-anchor recovery and anchor-guided sentence reconstruction. Stage 1 uses contrastive learning to align word-level EEG evidence with a fixed keyword vocabulary and recover ordere

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Robust Explanations for User Trust in Enterprise NLP Systems

arXiv:2604.12069v4 Announce Type: replace Abstract: Robust explanations are increasingly required for user trust in enterprise NLP, yet pre-deployment validation is difficult in the common case of black-box deployment (API-only access) where representation-based explainers are infeasible and existing studies provide limited guidance on whether explanations remain stable under real user noise, especially when organizations migrate from encoder classifiers to decoder LLMs. To close this gap, we propose a unified black-box robustness evaluation framework for token-level explanations based on leave-one-out occlusion, and operationalize explanation robustness with top-token flip rate under realistic perturbations (swap, deletion, shuffling, and back-translation) at multiple severity levels. Using this protocol, we conduct a systematic cross-architecture comparison across three benchmark datasets and six models spanning encoder and decoder families (BERT, RoBERTa, Qwen 7B/14B, Llama 8B/70B;

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

When Perplexity Lies: Generation-Focused Distillation of Hybrid Sequence Models

arXiv:2603.26556v2 Announce Type: replace Abstract: Converting a pretrained Transformer into a more efficient hybrid model through distillation offers a promising approach to reducing inference costs. However, achieving high-quality generation in distilled models requires careful joint design of both the student architecture and the distillation process. Many prior distillation works evaluate downstream multiple-choice benchmarks by ranking candidate answers with log-likelihood rather than requiring autoregressive generation, which can obscure important differences in model quality. For example, on overlapping benchmarks, we show that a 7B distilled model that nearly matches its teacher to within 0.2 pp under log-likelihood scoring falls behind by 20.8 pp when it must generate answers autoregressively. We investigate this phenomenon with GenDistill, a multi-stage pipeline we designed for distilling a pretrained Transformer into an efficient Hybrid Kimi Delta Attention (Hybrid-KDA) stud

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

KDFlow: A User-Friendly and Efficient Knowledge Distillation Framework for Large Language Models

arXiv:2603.01875v3 Announce Type: replace Abstract: Knowledge distillation (KD) is an essential technique to compress large language models (LLMs) into smaller ones. However, despite the distinct roles of the student model and the teacher model in KD, most existing frameworks still use a homogeneous training backend (e.g., FSDP and DeepSpeed) for both models, leading to suboptimal training efficiency. In this paper, we present a novel framework for LLM distillation, termed \textbf{KDFlow}, which features a decoupled architecture and employs SGLang for teacher inference. By bridging the training efficiency of FSDP2 and the inference efficiency of SGLang, KDFlow achieves full utilization of both advantages in a unified system. Moreover, instead of transferring full logits across different processes, our framework only transmits the teacher's hidden states using zero-copy data transfer and recomputes the logits on the student side, effectively balancing the communication cost and KD perfo

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Speculative Decoding with a Speculative Vocabulary

arXiv:2602.13836v2 Announce Type: replace Abstract: Speculative decoding has rapidly emerged as a leading approach for accelerating language model (LM) inference, as it offers substantial speedups while yielding identical outputs. This relies upon a small draft model, tasked with predicting the outputs of the target model. State-of-the-art speculative decoding methods use a draft model comprising a single decoder layer and output embedding matrix, with the latter dominating drafting time for the latest LMs. Recent work has sought to address this output distribution bottleneck by reducing the vocabulary of the draft model. While this can improve throughput, it compromises speculation effectiveness when the target token is out-of-vocabulary. In this paper, we argue for vocabulary speculation as an alternative to a reduced vocabulary. We propose SpecVocab, an efficient and effective method that selects a vocabulary subset per decoding step. Across a variety of tasks, we show that SpecVoca

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments

arXiv:2602.01244v3 Announce Type: replace Abstract: Training agentic models for terminal-based tasks critically depends on high-quality terminal trajectories that capture realistic long-horizon interactions across diverse domains. However, constructing such data at scale remains challenging due to two key requirements: \textbf{\emph{Executability}}, since each instance requires a suitable and often distinct Docker environment; and \textbf{\emph{Verifiability}}, because heterogeneous task outputs preclude unified, standardized verification. To address these challenges, we propose \textbf{TerminalTraj}, a scalable pipeline that (i) filters high-quality repositories to construct Dockerized execution environments, (ii) generates Docker-aligned task instances, and (iii) synthesizes agent trajectories with executable validation code. Using TerminalTraj, we curate 32K Docker images and generate 50,733 verified terminal trajectories across eight domains. Models trained on this data with the Qw

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Bifocal Attention: Harmonizing Geometric and Spectral Positional Embeddings for Algorithmic Generalization

arXiv:2601.22402v2 Announce Type: replace Abstract: Rotary Positional Embeddings (RoPE) have become the standard for Large Language Models (LLMs) due to their ability to encode relative positions through geometric rotation. However, we identify a significant limitation we term ''Spectral Rigidity'': standard RoPE utilizes a fixed geometric decay ($\theta^{-i}$) optimized for local syntactic coherence, which fails to capture the long-range, periodic structures inherent in recursive logic and algorithmic reasoning. This results in a ''Structure Gap'', where models trained on shallow reasoning chains fail to extrapolate to deeper recursive steps. In this work, we introduce Bifocal Attention, an architectural paradigm that decouples positional encoding into two distinct modalities: Geometric Eyes (Standard RoPE) for precise token-level manipulation, and Spectral Eyes (Learnable Harmonic Operators) for tracking long-range recursive depth. We propose a novel training protocol, Spectral Evolu

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Analysing Moral Bias in Finetuned LLMs through Mechanistic Interpretability

arXiv:2510.12229v3 Announce Type: replace Abstract: Large language models (LLMs) have been shown to internalize human-like biases during finetuning, yet the mechanisms by which these biases manifest remain unclear. In this work, we investigated whether the well-known Knobe effect, a moral bias in intentionality judgements, emerges in finetuned LLMs and whether it can be traced back to specific components of the model. We conducted a Layer-Patching analysis across 3 open-weights LLMs and demonstrated that the bias is not only learned during finetuning but also localized in a specific set of layers. Surprisingly, we found that patching activations from the corresponding pretrained model into just a few critical layers is sufficient to eliminate the effect. Our findings offer new evidence that social biases in LLMs can be interpreted, localized, and mitigated through targeted interventions, without the need for model retraining.

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Latent Fusion Jailbreak: Blending Harmful and Harmless Representations to Elicit Unsafe LLM Outputs

arXiv:2508.10029v3 Announce Type: replace Abstract: Safety-aligned large language models can still be manipulated through white-box interventions that modify their internal representations. We introduce Latent Fusion Jailbreak (LFJ), which works by pairing a harmful query with a structurally similar but benign counterpart, then interpolating their hidden states at carefully selected layers and token positions. Refusal-loss gradients determine exactly where to intervene, and we optimise layer-wise mixing coefficients using token-normalised compliance and refusal-suppression objectives. The edited prompt states propagate sequentially through the remaining transformer blocks. Across four safety benchmarks and five open-weight target models, LFJ reaches a macro-averaged attack success rate (ASR) of 94.13% under the white-box protocol we describe. Because LFJ directly accesses internal states, comparisons with prompt-only attacks serve as a descriptive reference rather than a matched evalua

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Learn to Memorize: Scalable Continual Learning in Semiparametric Models with Mixture-of-Neighbors Induction Memory

arXiv:2303.01421v2 Announce Type: replace Abstract: Semiparametric language models (LMs) have shown promise in various Natural Language Processing (NLP) tasks. However, they utilize non-parametric memory as static storage, which lacks learning capability and remains disconnected from the internal information flow of the parametric models, limiting scalability and efficiency. Based on recent interpretability theories of LMs, we reconceptualize the non-parametric memory represented by $k$NN-LM as a learnable Mixture-of-Neighbors Induction Memory (MoNIM), which synergizes the induction capabilities of attention heads with the memorization strength of feed-forward networks (FFN). By integrating into the model's information flow, MoNIM functions as an FFN-like bypass layer within the Transformer architecture, enabling effective learning of new knowledge. Extensive experiments demonstrate that MoNIM is a retentive and scalable continual learner in both data- and model-wise, enhancing the sca

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

An Exam for Active Observers

arXiv:2607.16165v1 Announce Type: cross Abstract: Human vision is a closed loop: gaze is continuously redirected by intermediate hypotheses rather than a single snapshot. Decades of psychophysics and cognitive science have argued that this active observation is essential for a wide range of tasks. Whether today's multimodal large language models (MLLMs) exercise active observation is an empirical question that current vision-language benchmarks do not answer. We introduce ActiveVision, a benchmark that makes active observation measurable for MLLMs, comprising 17 tasks across 3 categories. Tasks are designed to force repeated visual perception rather than a single static description. Frontier MLLMs collapse on ActiveVision: the highest-scoring model we evaluate, GPT-5.5 at the highest exposed reasoning-effort tier, solves only 10.6% of items and scores zero on 11 of the 17 tasks, and even Claude Fable 5, despite topping most reasoning and coding leaderboards, solves just 3.5%, far behin

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Understanding Reasoning from Pretraining to Post-Training

arXiv:2607.16097v1 Announce Type: cross Abstract: Reinforcement learning (RL) has become central to improving large language models (LLMs) on complex reasoning tasks, yet RL post-training is largely studied in isolation from the pretraining that precedes it. As a result, two basic questions remain open: (1) how do pretraining choices (model size, data) shape the returns to RL compute, and (2) what does RL actually do to the model? These questions are difficult to study in the standard LLM setting: pretraining corpora are vast and uncontrolled, making it hard to attribute behaviors to pretraining versus RL, and systematic compute sweeps across both stages are prohibitively expensive. To address these challenges, we use chess as a controlled testbed for studying reasoning across the full pretraining-to-post-training pipeline. We follow the standard LLM training pipeline by pretraining language models from 5M to 1B parameters on human chess games, supervised fine-tuning on synthetic reaso

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

HCIG: A Hierarchical Cross-Modal Incongruity Graph Network for Multimodal Sarcasm and Cyberbullying Detection

arXiv:2607.16076v1 Announce Type: cross Abstract: Multimodal sarcasm and cyberbullying detection remain challenging because the intended meaning often emerges from incongruity between textual and visual information rather than from either modality alone. Existing multimodal approaches primarily rely on feature fusion or cross-modal attention, which may not effectively capture hierarchical semantic inconsistencies across different levels of representation. To address this limitation, this paper proposes HCIG (Hierarchical Cross-modal Incongruity Graph Network), a novel framework that models cross-modal incongruity at token, phrase, and global levels using graph attention networks and adaptively integrates these representations through a learned hierarchical attention mechanism. As a complementary architecture, we also introduce GCCN (Graph-based Cross-modal Contradiction Network), which performs graph-based reasoning using contradiction-aware pooling for efficient multimodal interaction

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

CoWeaver: A Bi-directional, Learnable and Explainable Matching Engine for Mixed Human-Agent Science Collaboration

arXiv:2607.15545v1 Announce Type: cross Abstract: LLM-based agents excel at writing articles, coding and information retrieval. However, they fail to form strong collaborations within the scientific community due to the bidirectional, dynamic nature of the problem and a high demand of decision interpretability. We proposed COWEAVER, a bidirectional, learnable and explainable algorithm to match scientists and form strong collaborations within a human-agent network. COWEAVER matches candidates and requesters through filling capability gaps and filters candidates through a two-stage ranking step. Finally, the model explores newcomers by maintaining uncertainty-aware capability estimates and updating them through requester's feedback. We show that the selection mechanism of combining both exploration (UCB) and greedy of COWEAVER exceeds the greedy-only mechanism - the analytical best solution - on 6 out of the 20 tasks and performed on par with the greedy-only mechanism in terms of selecti

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Looped Latent Attention: Cross-Loop KV Compression for Looped Transformers

arXiv:2607.15456v1 Announce Type: cross Abstract: Looped, weight-tied Transformers reduce parameters by reusing a block, but decoding still stores a separate K/V cache for every recurrence step. We show that this loop-indexed cache is highly structured. For a fixed token, layer and head, K/V vectors trace a short low-rank trajectory across loops, while the head and layer axes remain much flatter. We introduce Looped Latent Attention (LLA), a post-training cache codec that stores compact K and V latents and reconstructs loop-specific K/V vectors only when attention reads them. The default per-head codec compresses recurrence, while LLA-2D also folds heads into one latent for the extreme-compression regime. The codec is initialized from SVD of teacher activations and refined with KL and attention-output distillation. At matched cache budget, per-head LLA outperforms head-axis MLA, cross-layer sharing, KV quantization and final-loop reuse, showing that the recurrent cache is low-rank but

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Hidden in Thought: Transferable Chain-of-Thought Artifacts Induce Harmful Behavior

arXiv:2607.15286v1 Announce Type: cross Abstract: We investigate whether harmful chain-of-thought (CoT) traces from compromised language models can transfer unsafe behaviour and be distilled into reusable jailbreak attacks. Using an emergent-misalignment organism and a refusal-ablated jailbroken organism, we transplant harmful CoTs into $29$ open-source and $5$ closed-source targets. Transferred traces raise harmful-response rates above $80\%$ on the most vulnerable open-source models, while semantically mismatched CoTs fail entirely. LLooM concept mining identifies four recurring components of harmful reasoning: proceduralisation, ethical decoupling, evasion, and target--vulnerability framing. Distilling these patterns into reusable system prompts produces effective black-box jailbreaks, outperforming direct CoT transplantation on strongly aligned models by up to an order of magnitude, including a $10\times$ improvement on GPT-4.1 AdvBench. Reasoning-enabled models are more than twice

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

ToolSciVer: Multimodal Scientific Claim Verification with Visual Tool Augmented Reinforcement Learning

arXiv:2607.16131v1 Announce Type: new Abstract: Multimodal Scientific Claim Verification (MSCV) requires models to verify scientific claims using visually grounded evidence from papers, including figures, tables, charts, and textual context. However, existing methods often fail because they struggle to locate decisive visual evidence, accurately read structured scientific visuals, and integrate multimodal observations into reliable reasoning. We introduce ToolSciVer, the first tool-augmented framework for MSCV to our knowledge. ToolSciVer equips a VLM with three type-aware visual tools, table row/column focus, chart-to-structure parsing, and high-resolution region zoom, which convert dense scientific visuals into explicit, claim-facing evidence, and trains the policy with Group Relative Policy Optimization (GRPO) under a composite reward of answer correctness, format validity, length control, tool-use efficiency, and tool-validity penalties. Experiments on SciVer and MuSciClaims datase

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Rate-Utility Frontiers for Language Encodings: Comparing Tokens, Bytes, and Pixels Under Controlled Linguistic Content

arXiv:2607.16117v1 Announce Type: new Abstract: Language models encode text as subword tokens, raw bytes, or rendered pixels, but these encodings are usually compared under modeling constraints that expose different amounts of linguistic content to models across different languages. We instead ask what each encoding preserves when both the content and the downstream capacity are controlled. Using verified parallel sentences across thirteen languages and five scripts, we compare tokens, bytes, and pixels through a shared bottleneck whose width is swept to trace rate-utility frontiers. This separates three quantities that are often conflated: the number of input positions an encoding creates, the latent capacity available after encoding, and the task-relevant information that survives compression. We evaluate three utilities: surface form preservation, cross-lingual sentence alignment, and topic classification. No encoding dominates across tasks or capacity regimes. Pixels preserve surfa

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Controlling Implicit Shortcut Reliance in L2 Spoken English Auto-markers

arXiv:2607.16085v1 Announce Type: new Abstract: Increasingly, speech and language processing tasks take either audio or text directly rather than extracting features from these as the input to the classifier or regressor. Often these systems make use of complex, for example transformer-based, processes that have the ability to derive highly non-linear mappings between the input and the output. Unfortunately these systems can also learn ''shortcuts'' where the classifier is overly reliant on particular aspects of the input to yield the output. For the task of language proficiency assessment, this over-reliance can enable learners to increase their score by exploiting the shortcut rather than improving their ability. This paper introduces a novel training criterion that is able to reduce the classifier's reliance on shortcuts, thus for example limiting this option for malpractice in language assessment. This process is illustrated on two forms of assessment system, one based on the audio

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Frontier Language Models Struggle to Copy: Text Can Be Better Viewed in 2D

arXiv:2607.16072v1 Announce Type: new Abstract: While large language models (LLMs) can solve advanced reasoning problems in seconds, we show that even frontier models fail to perform a much simpler operation: exactly copying an input string that lies well within their context windows. We attribute this failure to positional encodings in Transformer architectures, whose inductive bias favors copying through a shortcut based on matching local contexts rather than carefully locating the corresponding input positions. To address this issue, we introduce 2D-RoPE, which organizes text into a 2D grid rather than a 1D sequence and assigns each token a row ID and a column ID. Under this view, copying becomes simply retrieving input tokens at a fixed column offset, which makes the task easy to learn. In synthetic copy experiments, shallow Transformers with 2D-RoPE achieve perfect copying at input lengths hundreds of times longer than those seen during training, whereas standard positional encodi

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Frontier AI performance across the business disciplines: a case-grounded benchmark of knowledge work and analytical reasoning

arXiv:2607.16057v1 Announce Type: new Abstract: Large language models (LLMs) are improving rapidly as reflected in benchmark scores, yet these AI benchmarks largely test capabilities such as factual recall, narrow question answering, mathematical problem-solving, and coding and agentic tool-use. What remains poorly measured is AI progress on the analytical knowledge work white-collar professionals perform daily, including synthesizing complex information, exercising judgment under uncertainty and incomplete information, applying strategic and adversarial thinking in multi-stakeholder settings, weighing trade-offs, and producing defensible, structured analyses. This gap is even more pronounced for subjective components of such work, where success can be challenging to define. The "case method" form of education practiced by top business schools provides a natural foundation for addressing this measurement gap, and we construct BusinessCaseBench, a benchmark spanning hundreds of question

Source ↗
technology Mon, 20 Jul 2026 00:00:00 -0400
arXiv cs.CL

Loop the Loopies!

arXiv:2607.16051v1 Announce Type: new Abstract: We present Loopie, the most powerful looped Transformer to date. The Loopie series consists of two Mixture-of-Experts (MoE) models: a 20B-parameter model with 2B active parameters and a 6Bparameter model with 0.6B active parameters. Looped Transformers have long faced a challenge: given an N-fold increase in pre-training compute, increasing the parameter count by a factor of N usually outperforms looping a model N times. Loopie addresses this challenge. Extensive ablation studies, including comparisons with a vanilla 30B-A3B model, show that Loopie substantially outperforms vanilla Transformer baselines trained with the same compute budget. Our novel post-training pipeline equips Loopie with strong reasoning abilities. At the 2025 IMO and IPhO, Loopie achieves gold-medal performance without tools.

Source ↗
Showing 8101–8150 of 11029 signals
← Prev Page 163 of 221 Next →