Machine learning news — deep learning, reinforcement learning, neural architecture search, diffusion models, and new ML frameworks and libraries.
Ollama released version 0.30.0-rc23, a pre-release that shifts the software's architecture to directly support llama.cpp instead of building on top of GGML, and adds compatibility with the GGUF file format. The update al…
A new open-source tool, Auto GPU Kernel, autonomously discovers and optimizes GPU kernels, achieving a 34.93x average speedup to rank first in the DeepSeek Sparse Attention track of the MLSys 2026 FlashInfer AI Kernel Ge…
Researchers introduced Quantum Frog, a two-player cooperative game with a quantized-time mechanic where the environment advances only when a player acts. Using reinforcement learning, they found that a "rush strategy" of…
Researchers at arXiv have introduced "infilling extraction," a new method for extracting training data from diffusion language models (DLMs) that uses arbitrary binary masks instead of relying solely on prefix-conditione…
Researchers have identified three systematic failure modes in Audio LLMs when transcribing English-Mandarin code-switching speech: language omission, translation-instead-of-transcription, and hallucination. By applying D…
Researchers have released Nano World Models, a minimalist codebase for future video prediction that uses diffusion forcing to enable controlled studies of world-modeling components. The open-source platform provides a un…
Researchers have developed SkySeg, a collaborative framework enabling multiple unmanned aerial vehicles (UAVs) to perform real-time semantic segmentation onboard despite hardware constraints and environmental data shifts…
Researchers have introduced KREPE, the first generative representation learning method for hyper-relational knowledge graphs (HKGs) that can generate valid facts from arbitrarily masked queries, including completing part…
A new AI-driven workflow that uses detailed "constitutions" to define labeling categories and a frontier LLM to interpret them has reduced cross-model inconsistency by up to 57 times compared to standard paragraph defini…
Researchers have developed a gaze-aligned artificial intelligence framework that predicts perceived cycling safety by integrating human eye-tracking data into image analysis. The model, called EG-PCS, uses vision transfo…
A new study models the interaction between users and AI recommendation systems, finding that the optimal number of product suggestions depends on the cost of user communication versus search effort. Researchers analyzed …
Researchers have developed DRIVE, a dual-level skill modeling framework that separates web agent knowledge into natural language reasoning skills for transferable task logic and programmatic interaction skills for execut…
Researchers have developed a new method for selecting which large language models (LLMs) to include in multi-AI ensembles, treating the process as a combinatorial selection problem focused on model complementarity rather…
Researchers have developed EchoDistill, a self-distillation framework that improves Audio Large Language Models' resistance to real-world noise by aligning noisy student responses with clean semantic references from a fr…
Researchers have developed Diff-Instruct with Diffused Reward (DIDR), a data-free trajectory-level alignment framework that propagates reward-tilted clean-image distributions across all noise levels to improve one-step t…
Researchers have developed a new online weighted aggregation mechanism for mobile crowdsourcing platforms that ensures truthful feedback from workers during LLM fine-tuning. The mechanism, which dynamically adjusts worke…
Researchers introduced BoxLitE, a knowledge base embedding model that uses convex optimization to faithfully represent both factual data and conceptual hierarchies in DL-Lite$^{\mathcal{H}}$ ontologies. The model maps co…
Researchers have introduced IVR-R1, a reinforcement learning framework that iteratively realigns visual reasoning trajectories to correct errors in multimodal large language models. The method uses a reward-driven screen…
Researchers have developed EMMA, a physics-informed multimodal framework that recovers all identifiable dynamical parameters of a system directly from raw video, audio, and image-based time-series observations. The frame…
Researchers introduced Verifiable Transformers, a framework that converts localized circuits inside Transformer models into bounded, solver-checkable claims using SMT solvers. The approach directly verifies properties li…