We propose a new algorithm for fine-tuning large language models using reinforcement learning. Tapered Off-Policy REINFORCE (TOPR) uses an asymmetric, tapered variant of importance sampling to speed up learning while maintaining stable learning dynamics, even without the use of KL regularization. TOPR can be applied in a fully offline fashion, allows the handling of positive and negative examples in a unified framework, and benefits from the implementational simplicity that is typical of Monte Carlo algorithms. We demonstrate the effectiveness of our approach with a series of experiments on the GSM8K and MATH reasoning benchmarks, finding performance gains for training both a model for solution generation and as a generative verifier. We show that properly leveraging positive and negative examples alike in the off-policy regime simultaneously increases test-time accuracy and training data efficiency, all the while avoiding the ``wasted inference'' that comes with discarding negative examples. We find that this advantage persists over multiple iterations of training and can be amplified by dataset curation techniques, enabling us to match 70B-parameter model performance with 8B language models. As a corollary to this work, we find that REINFORCE's baseline parameter plays an important and unexpected role in defining dataset composition in the presence of negative examples, and is consequently critical in driving off-policy performance.

Current AI advances largely rely on scaling neural models and expanding training datasets to achieve generalization and robustness.

The recent success of generative AI highlights the crucial role of high-quality human feedback in building trustworthy AI systems. However, the increasing use of large language models (LLMs) by crowdsourcing workers poses a significant challenge: datasets intended to reflect human input may be compromised by LLM-generated responses. Existing LLM detection approaches often rely on high-dimensional training data such as text, making them unsuitable for structured annotation tasks like multiple-choice labeling. In this work, we investigate the potential of peer prediction --- a mechanism that evaluates the information within workers' responses --- to mitigate LLM-assisted cheating in crowdsourcing with a focus on annotation tasks.

Sharpness-aware minimization (SAM) has emerged as a highly effective technique to improve model generalization, but its underlying principles are not fully understood. We investigate m-sharpness, where SAM performance improves monotonically as the micro-batch size for computing perturbations decreases, a phenomenon critical for distributed training yet lacking rigorous explanation. We leverage an extended Stochastic Differential Equation (SDE) framework and analyze stochastic gradient noise (SGN) to characterize the dynamics of SAM variants, including n-SAM and m-SAM. Our analysis reveals that stochastic perturbations induce an implicit variance-based sharpness regularization whose strength increases as m decreases. Motivated by this insight, we propose Reweighted SAM (RW-SAM), which employs sharpness-weighted sampling to mimic the generalization benefits of m-SAM while remaining parallelizable.

Initialization plays a critical role in Deep Neural Network training, directly influencing convergence, stability, and generalization. Common approaches such as Glorot and He initializations rely on randomness, which can produce uneven weight distributions across layer connections. In this paper, we introduce the Sinusoidal initialization, a novel deterministic method that employs sinusoidal functions to construct structured weight matrices expressly to improve the spread and balance of weights throughout the network while simultaneously fostering a more uniform, well conditioned distribution of neuron activation states from the very first forward pass. Because Sinusoidal initialization begins with weights and activations that are already evenly and efficiently utilized, it delivers consistently faster convergence, greater training stability, and higher final accuracy across a wide range of models, including convolutional neural networks, vision transformers, and large language models. On average, our experiments show an increase of 4.8 % in final validation accuracy and 20.9 % in convergence speed. By replacing randomness with structure, this initialization provides a stronger and more reliable foundation for Deep Learning systems.

Large Language Models (LLMs) offer substantial promise for clinical natural language processing (NLP); however, a lack of standardized benchmarking methodologies limits their objective evaluation and practical translation. To address this gap, we introduce ClinBench, an open-source, multi-model, multi-domain benchmarking framework. ClinBench is designed for the rigorous evaluation of LLMs on important structured information extraction tasks (e.g., tumor staging, histologic diagnoses, atrial fibrillation, and social determinants of health) from unstructured clinical notes. The framework standardizes the evaluation pipeline by: (i) operating on consistently structured input datasets; (ii) employing dynamic, YAML-based prompting for uniform task definition; and (iii) enforcing output validation via JSON schemas, supporting robust comparison across diverse LLM architectures. We demonstrate ClinBench through a large-scale study of 11 prominent LLMs (e.g., GPT-4o series, LLaMA3 variants, Mixtral) across three clinical domains using configurations of public datasets (TCGA for lung cancer, MIMIC-IV-ECG for atrial fibrillation, and MIMIC notes for SDOH). Our results reveal significant performance-efficiency trade-offs. For example, when averaged across the four benchmarked clinical extraction tasks, GPT-3.5-turbo

Concept Bottleneck Models (CBMs) are interpretable machine learning models that ground their predictions on human-understandable concepts, allowing for targeted interventions in their decision-making process. However, when intervened on, CBMs assume the availability of humans that can identify the need to intervene and always provide correct interventions. Both assumptions are unrealistic and impractical, considering labor costs and human error-proneness. In contrast, Learning to Defer (L2D) extends supervised learning by allowing machine learning models to identify cases where a human is more likely to be correct than the model, thus leading to deferring systems with improved performance. In this work, we gain inspiration from L2D and propose Deferring CBMs (DCBMs), a novel framework that allows CBMs to learn when an intervention is needed. To this end, we model DCBMs as a composition of deferring systems and derive a consistent L2D loss to train them. Moreover, by relying on a CBM architecture, DCBMs can explain the reasons for deferring on the final task. Our results show that DCBMs can achieve high predictive performance and interpretability by deferring only when needed.

Recent advances in scene understanding have leveraged multimodal large language models (MLLMs) for 3D reasoning by capitalizing on their strong 2D pretraining.

France-Germany jet plans crash: Can Europe end reliance on US for security? France and Germany have announced this week that they are ditching a landmark project to jointly develop a sixth-generation fighter jet. French President Emmanuel Macron confirmed on Monday that the project is being terminated, in what is being seen as a major blow to efforts to boost defence cooperation between European Union states, a key issue amid uncertainty cast by United States President Donald Trump over the readiness of the US to help defend its NATO allies. Since 2019, the US president has been flirting with the idea of obtaining Greenland . His remarks about his desire for the island, a self-governing territory which is part of the Kingdom of Denmark, built to a crescendo at the start of this year, with European leaders signalling their displeasure with the idea and Trump even threatening additional trade tariffs on those countries standing in his way.

Planning with pretrained diffusion models has emerged as a promising approach for solving test-time guided control problems. Standard gradient guidance typically performs optimally under convex, differentiable reward landscapes. However, it shows substantially reduced effectiveness in real-world scenarios with non-convex objectives, non-differentiable constraints, and multi-reward structures. Furthermore, recent supervised planning approaches require task-specific training or value estimators, which limits test-time flexibility and zero-shot generalization. We propose a Tree-guided Diffusion Planner (TDP), a zero-shot test-time planning framework that balances exploration and exploitation through structured trajectory generation. We frame test-time planning as a tree search problem using a bi-level sampling process: (1) diverse parent trajectories are produced via training-free particle guidance to encourage broad exploration, and (2) sub-trajectories are refined through fast conditional denoising guided by task objectives. TDP addresses the limitations of gradient guidance by exploring diverse trajectory regions and harnessing gradient information across this expanded solution space using only pretrained models and test-time reward signals. We evaluate TDP on three diverse tasks: maze gold-picking, robot arm block manipulation, and AntMaze multi-goal exploration. TDP consistently outperforms state-of-the-art approaches on all tasks.