How Reinforcement Learning After Next-Token Prediction Facilitates Learning

A novel framework named SynthSeg Agents has been introduced for Zero Shot Weakly Supervised Semantic Segmentation (ZSWSSS), which generates synthetic training data without relying on real images. This approach utilizes two key modules: a Self Refine Prompt Agent that creates diverse image prompts and an Image Generation Agent that produces images based on these prompts, enhancing the capabilities of semantic segmentation tasks.

A novel framework named Denser has been introduced to enhance the efficiency of Large Language Models (LLMs) by optimizing information density separately for reasoning and answering phases. This dual-density inference approach allows for the use of compressed, symbol-rich language during intermediate computations while ensuring that final outputs remain human-readable.

The introduction of 3DLLM-Mem marks a significant advancement in the capabilities of Large Language Models (LLMs) by integrating long-term spatial-temporal memory for enhanced reasoning in dynamic 3D environments. This model is evaluated using the 3DMem-Bench, which includes over 26,000 trajectories and 2,892 tasks designed to test memory utilization in complex scenarios.

RecTok has been introduced as a novel approach to enhance high-dimensional visual tokenizers in diffusion models, addressing the inherent trade-off between dimensionality and generation quality. By employing flow semantic distillation and reconstruction-alignment distillation, RecTok aims to improve the semantic richness of the forward flow used in training diffusion transformers.

A comprehensive survey has been conducted on event-based mobile sensing, highlighting its evolution from 2014 to 2025. The study emphasizes the challenges posed by high data volume, noise, and the need for low-latency processing in mobile applications, particularly in the context of event cameras that offer high temporal resolution.

A recent study published on arXiv explores how low-level bitwise perturbations, or fault injections, in large language models (LLMs) can affect the semantic meaning of generated image captions while maintaining grammatical integrity. This research highlights the vulnerability of transformers to subtle hardware bit flips, which can significantly alter the narratives produced by AI systems.

A new approach called Inference Time Feature Injection has been introduced to enhance real-time recommendation systems in long-form video streaming. This method allows for the selective injection of recent user watch history at inference time, overcoming the limitations of static user features that are updated only daily. The technique has shown a statistically significant increase in user engagement metrics by 0.47%.

A new method has been proposed for efficiently determining the regularization parameter for low-rank MMSE filters using a Kronecker-product representation. This approach highlights the importance of selecting the correct regularization parameter, which is closely tied to rank selection, and demonstrates significant improvements over traditional methods through simulations.