arXiv AI

arXiv:2610.22599v1 Announce Type: cross Abstract: Recent advances in AI make Recursive Self-Improvement (RSI) increasingly plausible. Large Language Models (LLMs) can already propose, select, and run Machine Learning (ML) experiments. However, the space of possible design choices in ML is vast. An LLM that proposes an experiment needs a reliable intuition of whether it is worth running given the timing constraints of long training runs. In this work, we measure such intuition in current frontier LLMs by evaluating how accurately they predict the impact of self-generated ML experiment ideas before these are put into training runs. To this end, we construct a dataset of 288 LLM-proposed ideas on eight MLE-bench tasks. We then prompt five frontier LLMs (GPT-6 Astra, Claude Opus 5.5, Claude Fable 5.1, Kimi K3, and GLM-5.3) to predict the impact of each idea and compare their predictions with the actual outcomes after training. We report four main findings. First, LLM predictions are substantially better than random chance but far from accurate: even the best model, Claude Opus 5.5, explains only half of the variance in outcomes (R^2 = 0.49). Second, we notice that all tested models systematically underestimate the magnitude of the changes. Third, prediction quality varies widely across tasks. Finally, the choice of the coding harness that actually implements the ideas has little effect on outcomes, and we find no evidence of self-preference. Higher reasoning settings and additional context lead to modest gains. Overall, our results show that current LLMs have useful but limited intuition about which ML ideas will work. We release our dataset to support further study.

arXiv:2610.22586v1 Announce Type: cross Abstract: We study explanation fairness in hate speech detection. High classification performance does not always mean that a model produces equally plausible explanations for different demographic groups. Using the HateXplain dataset, we introduce Plausibility-EFG, a metric that measures the largest difference in explanation plausibility across demographic groups by comparing model-generated token attributions with human-annotated rationales. We evaluate BERT, RoBERTa, and ADV-RoBERTa, an adversarially trained model, using Integrated Gradients and SHAP. Our results show that ADV-RoBERTa improves classification performance and alignment between explanations and human rationales while reducing explanation disparities across demographic groups. These improvements are observed with both attribution methods and remain consistent across training, validation, and test splits. We also show that reducing prediction-level bias alone does not eliminate disparities in explanations, suggesting that predictive fairness and explanation fairness are related but distinct objectives. Finally, our fairness-performance analysis shows that stronger adversarial regularization can further reduce Plausibility-EFG, although it may cause a small decrease in macro-F1. Overall, this work presents explanation fairness as an important objective in hate speech moderation and introduces Fair-REF, a framework designed to reduce disparities in model explanations across demographic groups.

arXiv:2610.22511v1 Announce Type: cross Abstract: Evaluating language models and agents requires combining benchmark accuracy, judge-based assessment, runtime performance, tool-use behavior, and retrieval quality. Differences in interfaces, metric semantics, and trace formats complicate these comparisons and can hide failures in agentic workflows. We present AuditKIT, a source-available evaluation library that records these measurements in one result representation, with direction-aware metric contracts, configuration fingerprints, and extensible metrics and backends. For agents, it provides turn-aware tool-call matching, separates parallel from sequential execution, and distinguishes verified task outcomes from trajectory-level diagnostics. Missing or unobservable evidence is marked unscored rather than counted as failure. For retrieval-augmented generation, it combines ranked retrieval metrics, judged grounding, and deterministic evidence checks. Seven case studies spanning model comparison, serving runtimes, retrieval, tool calling, and interoperability show how answer extraction, runtime behavior, and metric selection change evaluation outcomes.

arXiv:2610.22484v1 Announce Type: cross Abstract: The input space of single-spike neural networks decomposes into causal pieces, and the number of causal pieces has been shown to provide a lower bound on the approximation error for some spiking neural networks. While this decomposition is applicable to a wide range of spiking neuron models, the limits of how many causal pieces can be realised by spiking neural networks are still unknown. In this work, we derive two complementary upper bounds on the number of causal pieces in feedforward single-spike neural networks. First, we establish a combinatorial bound given by the poly-Bernoulli numbers, which applies to a broad class of spiking neuron models. The bound is of order $e^{2(L-1)N\log N+\mathcal{O}(N)}$, with network depth $L$ and maximum layer width $N$. Second, for a specific set of neuron models, we derive a bound based on hyperplane arrangements. Applying the latter framework, we obtain bounds of order $\mathcal{O}\!\left(N^{(4L-3)(N_0-1)}\right)$ for the non-leaky integrate-and-fire model with exponential synapses and the ReLU-type spike response model, and $\mathcal{O}\!\left(N^{L(N_0-1)}\right)$ for the leaky integrate-and-fire model with Dirac-$\delta$ synapses, where $N_0$ is the input dimension. For the non-leaky integrate-and-fire model with Dirac-$\delta$ synapses, additional temporal structure allows us to sharpen the general combinatorial bound to one that depends only on $N_0$, independently of the network depth and widths.

arXiv:2610.22467v1 Announce Type: cross Abstract: Training models on their own outputs has attracted widespread attention as a possible route to advance AI capabilities. One prominent approach is self-distillation, where a model is trained to match the token-level predictions it would have produced when given access to additional context, such as a cue, feedback, or instruction. However, the safety properties of this training procedure are poorly understood. In this paper, we present a surprising safety failure of recursive self-distillation, where each self-distilled model becomes the teacher for the next round of training. We observe that when a safe model is recursively self-distilled on its own (safe) generations while receiving an additional non-harmful cue, it becomes increasingly misaligned across a diverse set of harmful queries; we call this Progressive Misalignment. We show this failure mode in both the Qwen3 and Phi-4 model families on SORRY-Bench and StrongREJECT. Progressive misalignment also occurs in a more realistic setup, where the fixed cue is replaced by feedback from a simulated user with benign intent. Using a gradient-based method, we are also able to find benign prompts that lead to progressive misalignment when applying recursive self-distillation, showing that harmful training prompts are not necessary for this failure mode to emerge. We hope our work encourages future investigations of the safety properties of AI systems recursively trained on their own data.

arXiv:2610.22464v1 Announce Type: cross Abstract: Social media platforms play a central role in shaping political discourse during elections, particularly in geopolitically contested regions. We study how political narratives emerged, aligned, and propagated during the 2024 Moldovan presidential election by analyzing publicly available posts collected from X, Facebook, and Reddit between August and October 2024. Combining multilingual stance detection, coordination analysis, and causal inference, we examine both descriptive and causal mechanisms underlying online political discourse. First, we apply large language models to classify stance toward European integration, Russia, and the two main presidential candidates, revealing clear cross-platform and cross-linguistic differences in how political positions are expressed. Second, we construct retweet-based and semantic alignment networks on X to identify amplification and aligned framing of dominating narratives. Finally, we use a proximal causal inference framework to estimate peer influence effects in stance formation. Our results indicate consistently positive and statistically significant average treatment effects across major stance categories. These findings provide evidence that amplification and peer exposure contribute to the reinforcement of political narratives online. This work offers the first analysis of social media discourse surrounding the 2024 Moldovan presidential election and demonstrates the value of integrating language-based annotation, network analysis, and causal inference for studying political influence in multilingual and geopolitically sensitive contexts.

arXiv:2610.22450v1 Announce Type: cross Abstract: Three dimensional modeling for autonomous driving must continuously estimate road geometry, obstacle occupancy, traffic participant behavior, and observational uncertainty under limited computational resources. Existing reconstruction methods can synthesize realistic novel views, but visual fidelity does not sufficiently characterize the geometric reliability required for driving. Generative world models can expand scenarios and predict future observations, yet may introduce content unsupported by measurements. We propose DriveWorld3D, a framework combining temporal video learning, measurement constraints from multiple sensors, and offline generative simulation. Its online component maintains a unified world state comprising road surfaces, volumetric occupancy, dynamic entities, and road relations, while separately representing physical occupancy, observability, and evidence provenance. Action conditioned future prediction and risk sensitive planning connect this representation to driving decisions. Its offline component reconstructs recorded scenes using neural radiance fields or 3D Gaussian Splatting, then expands long tail conditions using generative models while restricting the use of generated content as geometric ground truth. We specify data partitioning, sensor ablations, fault injection, paired closed loop testing, and real to simulation consistency assessment. The central hypothesis is that explicit evidence management can reduce dangerous false free space predictions and occlusion risk within a matched computational budget while retaining acceptable driving efficiency. All performance claims remain subject to experimental validation. Keywords: autonomous driving; 3D occupancy; sensor fusion; uncertainty; generative world models; NeRF; 3D Gaussian Splatting; closed loop simulation

arXiv:2610.22439v1 Announce Type: cross Abstract: Catastrophic forgetting remains a challenge in supervised continual learning, as adapting to new tasks causes representation drift and loss of previously learned information. This raises the question of which representational changes should be constrained while preserving plasticity. We therefore introduce Sparse Autoencoder Geometric Anchoring (SAGA), which converts a Sparse Autoencoder (SAE) trained at each task boundary into local, low-rank constraints on representation drift. Replay examples specify where the constraints apply, while leading right singular directions of the encoder Jacobian specify which changes are penalized. The resulting geometry is input dependent and can be retained without keeping the SAE for subsequent tasks' training. We establish its local relation to latent code matching, define its spectral truncation error, and derive a conditional bound on frozen-readout drift. With replay and snapshot history, the method achieves the highest mean Class-IL accuracy and lowest mean forgetting among the evaluated ER, DER++, ER-ACE, and SGD baselines on three pretrained ViT-B/16 benchmarks, with accuracy gains of 12.15, 4.79, and 3.22 percentage points over the strongest conventional baseline on CIFAR-100, CUB-200, and ImageNet-R, respectively. From scratch results are more mixed, while rank ablations and geometry controls show that SAE-induced low-rank geometry offers useful stability-plasticity trade-offs.

arXiv:2610.22435v1 Announce Type: cross Abstract: Distributed Transformer inference across co-located edge devices is fundamentally communication-bound, as exchanging activations between sequence-parallel partitions dominates inference latency under commodity edge bandwidths. Existing compression schemes reduce this cost but retain a structural bottleneck: the received data must still undergo key/value projection before attention can proceed, serializing computation behind communication. This paper presents Transformer Residual Activation Codebook Encoding (Trace), a framework that addresses these limitations. Trace places a Vector Quantizer (VQ) at the post-attention residual of every Transformer layer and precomputes codebook-indexed key/value representations. During inference, exchanged activations are encoded into compact codebook indices, allowing receivers to directly recover keys and values through table lookup instead of executing the downstream computations required to derive them. By eliminating the received data downstream computation, Trace enables communication--computation overlap and reduces a key latency bottleneck in distributed Transformer inference. A hierarchical clustering further structures the learned codebook into a multi-resolution hierarchy, enabling runtime adaptation of the communication bitrate without retraining. The framework supports two operating variants: one prioritizing accuracy and the other balancing accuracy and efficiency through buffer-compatible quantization. Experiments on ViT-Base (ImageNet-1K and CIFAR-100) and ViViT-B (UCF-101) show that Trace achieves up to 1.89% higher accuracy and up to 1.27x end-to-end inference speedup compared with the state-of-the-art VQ-based distributed inference framework. Compared with uncompressed sequence-parallel inference, Trace reduces communication overhead by up to 22,624x.

arXiv:2610.22425v1 Announce Type: cross Abstract: Safety-aligned language models are expected to refuse harmful requests while remaining useful for benign ones. However, because many evaluations score only the final response, they often conflate intent understanding with response-level safety behavior. To address this, we introduce RiskCueBench, a recoverability-controlled benchmark that separates these two stages by first measuring whether a model can successfully recover a high-level intent and then evaluating its safety behavior on the same prompt. RiskCueBench contains 160 latent intents and 640 abstract non-operational surface prompts across eight risk categories. The benchmark covers 25 completed model groups and 32,000 successful model outputs with final rubric-adjudicated labels. When conditioning on pairs where both surface forms are recoverable, we find that increased contextualized specificity induces a shift toward less safe behavior. Specifically, paired bootstrap estimates show that the refusal rate and global safety score decrease by 0.175 and 0.440, respectively. Furthermore, mixed effects modeling corroborates a 0.107 increase in unsafe cooperation. These results show that the safety behavior can be miscalibrated to the surface form even after the intent has been recovered, motivating safety evaluations that report response behavior conditional on semantic recoverability.

arXiv:2610.22423v1 Announce Type: cross Abstract: Deformable image registration is a fundamental challenge in medical image analysis. Recent deep learning-based methods typically adopt either one-pass architectures that complete registration in a single network iteration, or recursive architectures that refine registration through multiple passes. While recursive methods often improve accuracy, they incur substantial computational overhead and slower inference speed. In this paper, we challenge the necessity of recursion for high-performance deformable registration and propose M2CG, a one-pass registration network with Multi-level Multi-space Contextual Guidance. It performs multi-step progressive pyramid registration within a single iteration, guided at each step by contextual cues from three levels (previous, current, and subsequent steps) and four spaces (image space, image feature space, segmentation space, and segmentation feature space). Our M2CG supports both unsupervised (without any labels) and semi-supervised (with segmentation labels for partial training data) registration, and has achieved state-of-the-art performance across two well-benchmarked inter-/intra-patient medical registration tasks with eight public datasets. Our findings suggest that the current reliance on recursion may stem from the absence of sufficient contextual cues that comprehensively characterize the remaining distance to the target alignment.

arXiv:2610.22416v1 Announce Type: cross Abstract: Automatic melody harmonization is often evaluated through individual compositions or data-driven corpus models. This study examines whether an explainable harmonic decision model developed for individual songs can be applied to independently collected human annotations without changing its rules. We evaluate a frozen Explainable Melody Harmonizer (EMH) on selected HookTheory excerpts containing aligned melody and chord annotations. EMH generates chord sequences from melody alone using explicit criteria for melodic compatibility, harmonic function, persistence, cadence, meter-aware segmentation, and Viterbi-like optimization. Human chords are withheld during generation and used only for post-hoc comparison. Nineteen excerpts yielded 439 evaluable harmonic segments. Exact chord agreement reached 20.50%, chord-root agreement 31.66%, and harmonic-function agreement 47.62%, with functions assessed relative to the key inferred by EMH. Decision-level analysis identified 203 segments (46.24%) where the human chord lay outside the model's candidate vocabulary and 146 (33.26%) where it was available but not selected. These findings distinguish representational limitations from differences in chord selection without treating the human reference as the only musically valid solution. The study demonstrates a reproducible, interpretable approach to corpus-based evaluation of harmonic decision-making.

arXiv:2610.22407v1 Announce Type: cross Abstract: In existing agent-based simulations of multimodal urban mobility, agents, representing humans, choose their travel modes either using discrete choice or machine learning models fitted on structured travel survey data, or with rules explicitly written by the domain experts. Such approaches fix a priori the agents' possible behaviour space. Hence, a factor absent from variables or rules cannot affect agents' decisions. In reality, choosing a transport mode is a multidimensional decision in which objective and subjective factors intertwine and change over time. Integrating such complexity into a city-scale multimodal mobility simulation remains a challenge. Generative agents (GAs) built on large language models promise to respond to this challenge because they carry decision heuristics that travel surveys do not record. However, to the best of our knowledge, aligning their decisions with real household travel surveys remains an open question. In this paper, we build an agent-based model of daily multimodal mobility for Toulouse (France). We compare the performance of fifteen decision-makers (simple baselines, models fitted on the survey, GAs with minimal and tuned prompts, and a typed classifier) on the real household travel survey. Our results show that GAs with tuned prompts come close to the fitted models on the modal split of a simulated day without transport disruptions, but not trip by trip. GAs also react when unforeseen events occur. The recently released Jev-1.13 TypeSafe classifier gives surprisingly great results, showing that generating output text is not necessary. The results presented in this paper would be of interest to transport policy makers and service providers looking to apply generative agent-based modelling for improving urban mobility services.

arXiv:2610.22382v1 Announce Type: cross Abstract: LLMs can generate Lean specifications from task descriptions, but those descriptions are often underspecified or ambiguous and may omit verifier-relevant edge cases, yielding imperfect formal specifications. We study whether an elicitation agent (EA) can improve verifier-checked Lean specifications by seeking clarification from a reference-grounded simulated developer before formalization. The agent decomposes the task, identifies unresolved decisions, asks bounded clarification questions, and uses the answers to write a refined natural-language specification that an LLM formalizer then translates into Lean. On 189 Verina Lean 4 specification-generation tasks, we compare single-shot generation with elicitation under question budgets of k=3 and k=10, allowing at most three or ten EA questions per task, across same-model and cross-model configurations. The same-model Claude Sonnet 4.6 pipeline has negative strict pass@1 estimates at both budgets, showing elicitation is not a uniform accuracy booster. In contrast, the Sonnet 4.6-to-GPT-5.6 and Sonnet 4.6-to-Gemini 3.8 Flash pipelines have positive strict estimates. The largest strict result reaches 65.6%, a +6.3-point change (+12 tasks), in the Sonnet 4.6-to-GPT-5.6 pipeline. Six of eight conditions have positive compile-conditional estimates, with a maximum cross-model difference of +4.9 points among compiling outputs. Treatment estimates also differ across the basic and advanced splits; the largest advanced-task change occurs in the Sonnet 4.6-to-GPT-5.6 configuration at k=10, from 42.0% to 50.6%. Failure analysis identifies postconditions as the largest source of degradation. These findings suggest future autoformalization systems should treat elicited natural-language specifications as intermediate artifacts and optimize jointly for semantic correctness among compiling outputs and reliable translation of clarified information into Lean.

arXiv:2610.22379v1 Announce Type: cross Abstract: Calibrating expensive engineering simulators requires tuning model parameters to match reference data, a process heavily constrained by computational costs. This study investigates whether Large Language Models (LLMs), conditioned on physical diagnostic descriptions of simulation errors, can propose informative parameter updates more efficiently than conventional derivative-free optimizers. We test this on a cylinder-wake Large-Eddy Simulation and a damped-oscillator inverse problem. Physics-labeled prompts are evaluated against four text ablations (removed, anonymized, padded, misdirected) and standard optimizers (Nelder-Mead, Bayesian optimization, CMA-ES, and random search). Results demonstrate a strong early semantic advantage: the first physics-informed LLM proposal improves the initial design in 23 of 25 trials. After five evaluations, the median calibration error is 21.9 times lower than physics-free controls. However, this advantage is transient - disappearing by fifteen evaluations - and inherent LLM decoder variability yields a 9.8-fold performance spread. To address this, we introduce a hybrid workflow that uses the LLM as a warm-start, passing accumulated evaluations to a conventional optimizer. This two-stage numerical continuation policy reaches target accuracy in just 9 simulator runs, outperforming the LLM alone (12 runs), Nelder-Mead (17 runs), and Bayesian optimization (>60 runs). Supported by sensitivity analysis validating the test cases, this hybrid physics-labeled LLM scheme proves highly effective as a calibration agent under strict evaluation budgets.

arXiv:2610.21016v1 Announce Type: cross Abstract: Large language model (LLM) agents for health can accumulate reusable textual skills from task feedback, such as calculation criteria, measurement time points, and formula units. Deciding which skills to retain is consequential: useful rules improve later answers, while faulty generalizations propagate errors. We propose CausalEvo, a method for causal skill self-evolution that estimates the relative effects of deliberately changed skill sets while holding the model weights, feedback cases, and rule count fixed. At each evolution step, a frozen LLM proposes candidate rules, which are pooled with the incumbent skills so that the current skill set remains a feasible choice. Paired skill replacements against shared background rules are scored on the same feedback cases; least squares converts the score differences into relative skill values, and the top-ranked rules are retained, favoring incumbents on ties. A decision-loss bound separates candidate support, approximation, and estimation errors. On six MedCalc-Bench-Verified tasks with 120 held-out cases, CausalEvo achieves 60.00% accuracy, compared with 55.83% for SkillOpt and 51.67% without skills. Across the recorded training configurations, which use different optimization budgets, it consumes 55.6% fewer training tokens than SkillOpt. Gains concentrate in particular calculations alongside regressions on others, and paired accuracy intervals include zero. CausalEvo turns task feedback into controlled decisions about skill retention and replacement. Project repository: https://github.com/txbrook/CausalEvo

arXiv:2610.23845v1 Announce Type: new Abstract: Reproducible and analytically robust research findings are essential to scientific progress, but evaluating these properties typically requires costly manual effort. We investigate whether LLM agents can help automate reproduction and robustness analysis, using a medical case study grounded in the Surveillance, Epidemiology, and End Results (SEER) database, a widely used U.S. cancer registry underlying tens of thousands of published studies. First, we introduce SEER-Repro, a benchmark of 20 published SEER-based papers together with the data and resources required to reproduce their findings. Second, we create an AI agent that successfully reproduces 77.4% of 2,190 atomic claims, while providing executable baselines for further analysis. Third, we develop an agentic analysis pipeline to test the robustness of the medical findings by identifying analytical choices underlying each finding, proposing defensible alternatives, and rerunning the analysis to determine whether those choices affect the conclusion. Across published papers in SEER-Repro, our pipeline identifies potentially consequential analytical choices in 9.5% of fully reproduced findings and flags them for expert review. These results show how LLM agents can support scientific review by reproducing findings, systematically testing their robustness, and directing expert attention toward potentially consequential analytical choices.

arXiv:2610.23826v1 Announce Type: new Abstract: Self-driving laboratories (SDLs) accelerate scientific discovery by automating experimental execution and optimization. AI agents can increasingly propose new experiments, but existing SDLs can only run the procedures they were built for. Adapting them to a new experiment still takes extensive manual engineering, which makes laboratory flexibility the bottleneck for scaling discovery. We introduce the self-building laboratory (SBL), which designs, configures, and operates its own laboratory setup given a high-level scientific goal. SBL starts from whatever already exists, from an empty room to a working SDL, and builds the missing capabilities: the lab layout, its digital twin, robot skills, and experimental workflows. We demonstrate the system working across different robots and setups, scaling from a single workstation to laboratories with several robots working together. We validate SBL's end-to-end capabilities through a real-world quantum-dot synthesis experiment. By shifting the role of human researchers from manually configuring and programming device automation to specifying scientific objectives, SBL makes scientific automation more accessible, flexible, and scalable. Project website: https://self-building-labs.github.io/

arXiv:2610.23801v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) explain foundation model representations by decomposing activations into sparse combinations of learned dictionary atoms. However, the learned dictionaries can vary across training runs, limiting the reliability and reproducibility of SAE-based analyses. We propose Consistent Sparse Dictionary Learning (ConSDL), which replaces the standard linear encoder with per-sample sparse coding. This is motivated by our synthetic analysis showing that coefficients from a single linear encoder pass are suboptimal compared with those obtained by per-sample optimization, suggesting that sparse coding is a reliable mechanism for dictionary learning. Across 16 synthetic settings, ConSDL achieves near-perfect dictionary and coefficient recovery. These gains translate to state-of-the-art cross-seed consistency across four vision and language settings, where ConSDL improves dictionary and coefficient consistency by 17.6% and 35.9%, respectively, over the average of six SAE baselines.

arXiv:2610.23796v1 Announce Type: new Abstract: Transformer prefill reuses weights across tokens and is dominated by matrix multiplication and activation data movement; single-token decode has limited reuse and is dominated by weight and KV-cache reads. We present FUTRANSFORMER, a CUDA implementation of a Llama-style Transformer block for NVIDIA Hopper GPUs. It applies CODA's GEMM-plus-epilogue approach to inference, adding direct KV-cache placement, folding the fixed RMSNorm gain into weights, writing FP8 output directly with delayed scaling, and supporting prefill without padding. Decode uses INT4 projection weights and an INT4 KV-cache copy. Weights and keys follow Tensor Core fragment layouts for direct dequantization into MMA operands without full intermediate tiles. A block-diagonal Hadamard transform reduces cache quantization error from outlier channels. On one H100, prefill is up to 1.48x faster than compiled PyTorch FP8 and reaches 1.29 PFLOP/s. For gate/up projection with SwiGLU at the 8B shape, our BF16 path is 1.13x faster than CODA. Our FP8 path is 2.02-2.18x faster than the evaluated BF16 CODA path. Across 8B and 70B shapes at batches 1-32, INT4 decode is 1.3-2.4x faster than the faster of PyTorch's BF16 and INT4 blocks at each point: 2.1-3.3x faster than compiled, CUDA-graphed BF16 and 1.3-6.3x faster than the same block using PyTorch's INT4 projection kernel, which targets a single query row and loses to BF16 at batch 32. It is also 31-51% faster than our FP8 path. Complete Llama-3.1-8B decode takes 2.8 ms per step at batch one, versus 8.1 ms for the compiled transformers implementation. Decoding every WikiText-2 token through the INT4 weight path raises perplexity from 6.24 to 7.23 with round-to-nearest weights or 6.99 with GPTQ in the same format; the rotated INT4 cache adds 0.20. The kernels match a quantized reference computation within 0.001 perplexity.