Large language models (LLMs) excel at structured, verifiable tasks, but their low-entropy bias can produce homogeneous and predictable outputs, limiting their utility for open-ended scientific ideation. Effective discovery, however, spans a broader creative spectrum: from structured day science to loosely structured, serendipitous night science that reaches ideas beyond those typically considered. We introduce AI Night-Scientist, an agentic framework that uses reinforcement learning to teach models when and how to depart from predictable reasoning. Grounded in cognitive science, we model creativity along three axes: action (what to do and how creatively), process (when to explore versus exploit), and outcome (the novelty and usefulness of the resulting idea). We use these axes to train models with GRPO, exposing them to varying degrees and forms of creativity throughout training. This produces substantially more diverse scientific proposals, expanding the range of research directions by 27.8% and contribution types by 14.9% over the base model. It also improves predicted citation impact by up to 32.0 percentage points and originality by 66.2 points. These gains cannot be reproduced by simply increasing decoding temperature; instead, we find that semantic guidance specifying what kind of creativity to pursue is critical. Overall, our results suggest that creativity is a learnable, multi-level ability that can be shaped to help researchers reach ideas beyond those typically explored by LLMs.
Перевод готовится · пока описание источникаГрафик появится после повторных замеров. Текущий показатель уже получен из источника.
2 голосов источникаТолько реальные замеры. История до подключения источника не восстанавливается.
В статье представлена GAGAR — фреймворк для распределения кредитов с учетом качества в RL-агентах для кода. Используется динамическая выборка и SFT-обученный агентный оценщик для ранжирования траекторий, проходящих тесты, и корректировки преимуществ для приоритизации более качественных реализаций. Метод был протестирован на промышленных кодовых агентах с большими количествами параметров.
3 дня назадDaily PapersVisual backbones have evolved from Convolutional Neural Networks (CNNs) with local aggregation to Vision Transformers (ViTs) with global interactions, State-Space Models (SSMs) with input-dependent state transitions, and Test-Time Training (TTT) layers that adapt an inner learner while processing an image. Across this progression, visual computation has become increasingly adaptive to each input, yet the rules governing that adaptation remain largely prescribed by the trained backbone. We introduce VisionHOPE, the first generic visual backbone formulated as a self-modifying learning system, in which what the model remembers and how it learns co-evolve within an image. Building on the self-referential construction of Nested Learning (NL), VisionHOPE realizes this co-evolution through five coupled memories that store content, generate key and value representations, and govern learning rate and retention. These memories evolve jointly as visual context accumulates along each scan. However, directly applying the unconstrained self-referential update to a visual backbone leads to instability. We therefore derive a stability-matched step-size control scheme that combines a soft cap on self-referential injection with a spectral clamp on the retained memory transition, and prove that the resulting memory dynamics are non-expansive along each scan. For two-dimensional feature maps, we adapt NL's chunk formulation by aligning chunks with image rows and columns across four directional scans. The proposed VisionHOPE achieves competitive results on ImageNet-1K, COCO, and ADE20K, establishing self-modifying learning systems as a practical foundation for general-purpose visual backbones. The code is available at https://github.com/PSRben/VisionHOPE.
позавчераDaily PapersВ статье исследуется линейность в больших языковых моделях (LLM), демонстрируя, что объединение входов из разных потоков текста приводит к суперпозиции распределений следующих токенов. Авторы предполагают, что эта линейность является врожденным свойством архитектуры Transformer и может быть восстановлена с помощью тонкой настройки, позволяя генерировать два согласованных продолжения за один проход.
5 дней назадDaily PapersВ этой статье представлена FuseReg, метод, который заменяет эвристическую фьюзинг-фьюзинг в представительных автокодерах (RAE) обучением на случайных подмножествах слоев кодировщика. Подход уменьшает разрыв между реконструкцией и генерацией, повышая устойчивость к выбору фьюзинга слоев, достигая более высокого PSNR и более низких значений FID генерации без изменения предобученного кодировщика.
4 дня назадDaily PapersVision-language (VL) pretraining using paired chest X-ray (CXR) images and radiology reports has shown strong potential for medical image understanding. However, existing methods often remain dependent on task-specific finetuning because radiology reports are lengthy, clinically dense, and difficult to align with simple zero-shot prompts. Recent sentence-level approaches partially address this limitation using clinical phrases extracted by large language models (LLMs), but they largely overlook the intrinsic characteristics of radiology discourse. In particular, limited positive-pair diversity constrains further gains, while clinically equivalent sentences frequently recur across patients, creating false negatives in contrastive learning. To address these issues, we propose SentZero, an enhanced sentence-centric VL pretraining framework for zero-shot, multi-task CXR analysis. SentZero introduces LLM-based abstract-level sentence structuring and mapping to expand positive-pair diversity, together with an additional loss term to mitigate false negatives. We further introduce sentence-conditioned residual modulation of visual embeddings, enabling visual features to adapt to the semantic characteristics of each input sentence. Across diverse downstream tasks and datasets, SentZero improves zero-shot generalization and outperforms prior multi-task zero-shot methods.
вчераDaily PapersВ этой статье представлена методика повышения способности больших языковых моделей (LLM) обучаться на основе контекста с использованием искажённых публичных документов. Подход включает переписывание документов, генерацию вопросов с элементами рассуждений и обучение студенческой модели на этих синтетических образцах, что приводит к улучшению производительности на CL-bench и более широкому применению в задачах, таких как понимание длинного контекста и рассуждения.
позавчера