Since the advent of large language models (LLMs), research has focused on instruction following and deductive reasoning. A central question remains: can these models discover new knowledge, and how can we evaluate this ability? We address this by studying abductive reasoning-the generation of plausible hypotheses to explain observations-and introduce GEAR (General Evaluation for Abductive Reasoning), a general-purpose, fully automated, transparent, and label-free evaluation paradigm. GEAR scores hypothesis sets by three metrics: consistency (each hypothesis explains the observations), generalizability (consistent hypotheses make meaningful predictions on unseen inputs), and diversity (the set covers distinct predictions and patterns). Built this way, GEAR is scalable (no human gold answers), reliable (deterministic scoring aligned with classical abduction), and open-ended (scores improve only when models produce new plausible hypotheses, unlike static benchmarks that saturate once accuracy is high). Using GEAR, we conduct a fine-grained study of nine LLMs on four abduction benchmarks with 1,500 problems, generating over 50,000 candidate hypotheses and revealing model differences obscured by gold-answer or purely human evaluations. We further propose a momentum-based curriculum that adjusts GEAR-derived training data by learning velocity: it starts with what the model learns quickly and shifts toward harder objectives such as generating diverse hypotheses once the model is confident on foundational objectives. Without gold-label supervision, this strategy improves all GEAR objectives and these gains transfer to established abductive reasoning benchmarks. Taken together, GEAR provides a principled framework that evaluates abduction and supplies label-free, scalable training signals that help LLMs produce more diverse and reliable hypotheses.

LLM leaderboards often rely on single stochastic runs, but how many repetitions are required for reliable conclusions remains unclear. We re-evaluate eight state-of-the-art models on the AI4Math Benchmark with three independent runs per setting. Using mixed-effects logistic regression, domain-level marginal means, rank-instability analysis, and run-to-run reliability, we assessed the value of additional repetitions. Our findings shows that Single-run leaderboards are brittle: 10/12 slices (83\%) invert at least one pairwise rank relative to the three-run majority, despite a zero sign-flip rate for pairwise significance and moderate overall interclass correlation. Averaging runs yields modest SE shrinkage ($\sim$5\% from one to three) but large ranking gains; two runs remove $\sim$83\% of single-run inversions. We provide cost-aware guidance for practitioners: treat evaluation as an experiment, report uncertainty, and use $\geq 2$ repetitions under stochastic decoding. These practices improve robustness while remaining feasible for small teams and help align model comparisons with real-world reliability.

Sequence classification is essential in NLP for understanding and categorizing language patterns in tasks like sentiment analysis, intent detection, and topic classification. Transformer-based models, despite achieving state-of-the-art performance, have inherent limitations due to quadratic time and memory complexity, restricting their input length. Although extensive efforts have aimed at reducing computational demands, processing extensive contexts remains challenging. To overcome these limitations, we propose ResFormer, a novel neural network architecture designed to model varying context lengths efficiently through a cascaded methodology. ResFormer integrates an reservoir computing network featuring a nonlinear readout to effectively capture long-term contextual dependencies in linear time. Concurrently, short-term dependencies within sentences are modeled using a conventional Transformer architecture with fixed-length inputs. Experiments demonstrate that ResFormer significantly outperforms baseline models of DeepSeek-Qwen and ModernBERT, delivering an accuracy improvement of up to +22.3% on the EmoryNLP dataset and consistent gains on MultiWOZ, MELD, and IEMOCAP. In addition, ResFormer exhibits reduced memory consumption, underscoring its effectiveness and efficiency in modeling extensive contextual information.

The advent of transformers has in recent years led to powerful and revolutionary Large Language Models (LLMs). Despite this, our understanding on the capability of transformers is still meager. In this invited contribution, we recount the rapid progress in the last few years to the question of what transformers can do. In particular, we will see the integral role of logic and automata (also with some help from circuit complexity) in answering this question. We also mention several open problems at the intersection of logic, automata, verification and transformers.

Neural scaling laws have played a central role in modern machine learning, driving the field's ever-expanding scaling of parameters, data and compute. While much research has gone into fitting scaling laws and predicting performance on pretraining losses and on discriminative evaluations such as multiple-choice question-answering benchmarks, comparatively little research has been done on fitting scaling laws and predicting performance on generative evaluations such as mathematical problem-solving or coding. In this work, we propose and evaluate three different pretraining scaling laws for fitting pass-at-k on generative evaluations and for predicting pass-at-k of the most expensive model using the performance of cheaper models. Our three scaling laws differ in the covariates used: (1) pretraining compute, (2) model parameters and pretraining tokens, (3) log likelihoods of gold reference solutions. We make four main contributions: First, we show how generative evaluations offer new hyperparameters (in our setting, k) that researchers can use to control the scaling laws parameters and the predictability of performance. Second, in terms of scaling law parameters, we find that the compute scaling law and parameters + tokens scaling law stabilize for the last 1.5 2.5 orders of magnitude, whereas the gold reference likelihood scaling law stabilizes for the last 5 orders of magnitude.

Diffusion language models (DLMs) have strong theoretical efficiency but are limited by fixed-length decoding and incompatibility with key-value (KV) caches. Block diffusion mitigates these issues, yet still enforces a fixed block size and requires expensive training. We introduce Next Sequence Prediction (NSP), which unifies next-token and next-block prediction, enabling the model to adaptively determine the generation length at each step. When the length is fixed to 1, NSP reduces to standard next-token prediction. Building on NSP, we propose Sequential Diffusion Language Model (SDLM), which can retrofit pre-trained autoregressive language models (ALMs) at minimal cost. Specifically, SDLM performs diffusion inference within fixed-size mask blocks, but dynamically decodes consecutive subsequences based on model confidence, thereby preserving KV-cache compatibility and improving robustness to varying uncertainty and semantics across the sequence. Experiments show that SDLM matches or surpasses strong autoregressive baselines using only 3.5M training samples, while achieving 2.1 higher throughput than Qwen-2.5. Notably, the SDLM-32B model delivers even more pronounced efficiency gains, demonstrating the strong scalability potential of our modeling paradigm. Project page and codes: https://github.com/OpenGVLab/SDLM

MCP standardizes how LLMs interact with external systems, forming the foundation for general agents. However, existing MCP benchmarks remain narrow in scope: they focus on read-heavy tasks or tasks with limited interaction depth, and fail to capture the complexity and realism of real-world workflows. To address this gap, we propose MCPMark, a benchmark designed to evaluate MCP use in a more realistic and comprehensive manner. It consists of $127$ high-quality tasks collaboratively created by domain experts and AI agents. Each task begins with a curated initial state and includes a programmatic script for automatic verification. These tasks demand richer and more diverse interactions with the environment, involving a broad range of create, read, update, and delete (CRUD) operations. We conduct a comprehensive evaluation of cutting-edge LLMs using a minimal agent framework that operates in a tool-calling loop. Empirical results show that the best-performing model, gpt-5-medium, reaches only $52.56$\% pass@1 and $33.86$\% pass^4, while other widely regarded strong models, including claude-sonnet-4 and o3, fall below $30$\% pass@1 and $15$\% pass^4. On average, LLMs require $16.2$ execution turns and $17.4$ tool calls per task, significantly surpassing those in previous MCP benchmarks and highlighting the stress-testing nature of MCPMark.

Learning to learn is becoming a science, driven by the convergence of knowledge tracing, signal processing, and generative AI to model student learning states and optimize education. We propose CoTutor, an AI-driven model that enhances Bayesian Knowledge Tracing with signal processing techniques to improve student progress modeling and deliver adaptive feedback and strategies. Deployed as an AI copilot, CoTutor combines generative AI with adaptive learning technology. In university trials, it has demonstrated measurable improvements in learning outcomes while outperforming conventional educational tools. Our results highlight its potential for AI-driven personalization, scalability, and future opportunities for advancing privacy and ethical considerations in educational technology. Inspired by Richard Hamming's vision of computer-aided 'learning to learn,' CoTutor applies convex optimization and signal processing to automate and scale up learning analytics, while reserving pedagogical judgment for humans, ensuring AI facilitates the process of knowledge tracing while enabling learners to uncover new insights.

Scientific discovery is often slowed by the manual development of computational tools needed to analyze complex experimental data. Building such tools is costly and time-consuming because scientists must iteratively review literature, test modeling and scientific assumptions against empirical data, and implement these insights into efficient software. Large language models (LLMs) have demonstrated strong capabilities in synthesizing literature, reasoning with empirical data, and generating domain-specific code, offering new opportunities to accelerate computational method development. Existing LLM-based systems either focus on performing scientific analyses using existing computational methods or on developing computational methods or models for general machine learning without effectively integrating the often unstructured knowledge specific to scientific domains. Here, we introduce TusoAI , an agentic AI system that takes a scientific task description with an evaluation function and autonomously develops and optimizes computational methods for the application. TusoAI integrates domain knowledge into a knowledge tree representation and performs iterative, domain-specific optimization and model diagnosis, improving performance over a pool of candidate solutions. We conducted comprehensive benchmark evaluations demonstrating that TusoAI outperforms state-of-the-art expert methods, MLE agents, and scientific AI agents across diverse tasks, such as single-cell RNA-seq data denoising and satellite-based earth monitoring. Applying TusoAI to two key open problems in genetics improved existing computational methods and uncovered novel biology, including 9 new associations between autoimmune diseases and T cell subtypes and 7 previously unreported links between disease variants linked to their target genes. Our code is publicly available at https://github.com/Alistair-Turcan/TusoAI.

Preference alignment is a critical step in making Large Language Models (LLMs) useful and aligned with (human) preferences. Existing approaches such as Reinforcement Learning from Human Feedback or Direct Preference Optimization typically require curated data and expensive optimization over billions of parameters, and eventually lead to persistent task-specific models. In this work, we introduce Preference alignment of Large Language Models via Residual Steering (PaLRS), a training-free method that exploits preference signals encoded in the residual streams of LLMs. From as few as one hundred preference pairs, PaLRS extracts lightweight, plug-and-play steering vectors that can be applied at inference time to push models toward preferred behaviors. We evaluate PaLRS on various small-to-medium-scale open-source LLMs, showing that PaLRS-aligned models achieve consistent gains on mathematical reasoning and code generation benchmarks while preserving baseline general-purpose performance. Moreover, when compared to DPO-aligned models, they perform better with huge time savings. Our findings highlight that PaLRS offers an effective, much more efficient and flexible alternative to standard preference optimization pipelines, offering a training-free, plug-and-play mechanism for alignment with minimal data.