Tag
This paper investigates the role of gating mechanisms in State Space Models, demonstrating that they promote memorization over in-context learning, yet can improve generalization to long sequences.
This paper challenges the distortion hypothesis for few-shot degradation in language models by introducing a random-text control, showing that representation shift is largely due to prompt length, and models with higher content delta benefit more from few-shot prompting.
This paper introduces GPT-Policy, a framework for in-context robot learning using vision-language models, enabling robots to learn from demonstrations without gradient updates. It evaluates the framework in real-robot trials, showing improved task completion.
A controlled study finds that feature engineering gains diminish for stronger tabular foundation models, while adding in-context information from related datasets still improves performance.
LimiX-2 is a pretrained foundation model for structured data that uses contextual mechanism networks to achieve #1 on major tabular benchmarks, supporting multiple tasks without task-specific parameter updates.
The article explores whether Astra represents a ChatGPT-like breakthrough in robotics, highlighting its in-context learning and reasoning capabilities through a Twitter thread.
The paper proposes the Convergent Emergence Hypothesis, stating that few-shot in-context learning emerges with a common cross-modality difficulty profile, and provides empirical support through experiments on six modalities, showing correlated effects in five.
A tweet claims that GPT-6 Astra demonstrates AGI by enabling a robot arm to perform novel physical tasks from human demonstrations on the first attempt.
The article discusses the GPT-6 Astra model's capability to perform in-context learning in physical world environments, representing a significant advancement in embodied AI.
This paper introduces a contrastive modeling framework for multimodal in-context learning to improve reasoning path alignment, enhancing performance on tasks like visual question answering.
This paper details the Eloquence team's approaches for Task 2 of the Interspeech 2026 MLC-SLM challenge, which involves multilingual multiple-choice question answering using speech LLMs with fine-tuning, in-context learning, and retrieval systems.
Skild AI's S1 model enables robots to learn from video demonstrations without retraining, achieving $100M ARR in 10 months and deploying across multiple industries.
MOAE proposes a Pareto-preserving evolutionary search method to simultaneously optimize multiple objectives like task performance, trajectory quality, and safety for LLM agents without fixed scalarization during search.
This paper evaluates retrieval-based in-context learning approaches for detecting criminally relevant hate speech in German social media posts, finding that few-shot prompting outperforms zero-shot but retrieval methods offer marginal gains, with models better suited for triage than autonomous moderation.
Supermemory highlights the importance of continual learning for AI agents and introduces learner-1, a new model designed to advance memory and in-context learning for various use cases.
This paper introduces causal foundation models (CFMs), which are pretrained neural networks that estimate causal quantities on new datasets using in-context learning without requiring fine-tuning.
Introduces Xiaomi-TabLDM, a tabular foundation model that leverages synthetic data and in-context learning for superior prediction accuracy without task-specific fine-tuning, achieving top rankings on multiple benchmarks.
This paper compares rule-based and example-based in-context learning in LLMs, finding that rules generally lead to more reliable learning across diverse tasks, with implications for model and task properties.
This paper introduces the Conformal Relevance framework, which uses in-context learning and ensembling to improve conciseness in NLP tasks while maintaining coverage guarantees via conformal prediction.
SALA is a Semantic-Aware Logical Alignment framework that improves demonstration selection for complex reasoning in in-context learning by automatically learning task-specific reasoning operations and using dynamic time warping for flexible alignment, outperforming existing methods.