Category
2,906 papers
As large language models (LLMs) scale rapidly, dense full-parameter adaptation becomes increasingly expensive, motivating sparse and modular architectures such as Mixture-of-Experts (MoE) models. This shift raises a key question for parameter-efficient fine-tuning (PEFT): at what granularity should parameters be select...
Results show that persistent program growth can move recurring control out of model context and into low-cost code, yielding reusable specialist agents that remain effective with smaller deployment models.
This work examines how reachable behavior, optimization budgets, and persistent adaptation shape exposure to proxy error, and develops a comparative framework for reward hacking across these three optimization substrates: weights, selection, and text.
Evaluated on two reasoning domains, math and multi-hop reasoning, across 1-8B models from different families, LoT consistently improves over KD and converges faster than staged curricula, highlighting the value of adaptive progression.
Reach audiences
Advertise in front of researchers, engineers, and readers.
This work builds four modules (for drug-discovery chemistry, materials science, machine learning, and mineral geochemistry) in the ArticleMiner framework, and evaluates them on 163 papers, including a new geochemistry benchmark with expert-curated ground truth.
This work treats attention as a routing map from one forward pass, keep a small set of routes that point toward the answer, and asks whether those routes actually matter, and whether those routes actually matter.
Analyzing neuroimaging data requires specialized coding and statistical expertise, which limits accessibility for researchers without computational backgrounds. We present the AI Neuroscientist, a language agent for interactive data exploration. The system integrates a large language model (LLM) with a neuroimaging too...
Our research topic has inspired studies on new results on lightweight computer vision (CV) models and edge deployment strategies for crop production. With the proliferation of unmanned aerial vehicles (UAVs), autonomous harvesters, and Internet-of-Things (IoT) devices in agriculture, the demand for efficient, accurate,...
Introduction The purpose of this study was to strengthen 5th grade elementary reading, math, and ESOL teachers' instructional expertise in integrating oral and written language development within Tier 1 differentiated instruction for multilingual language learners (MLLs), through a job-embedded professional development...
Background. The rapid development of digital technology in sports education has opened new opportunities for interactive learning and skill improvement. In floorball, traditional training methods often lack personalized guidance and real-time feedback, limiting athletes’ progress. Therefore, this study aims to develop...
Chemists read 1 H and 13 C NMR daily, and it is tempting to paste the peak lists into a large language model. Given only those two peak lists for 105 molecules balanced by molecular complexity and compound classes, six frontier LLMs (Claude Opus 4.8, DeepSeek V4 Pro, Gemini 3.1 Pro, GPT-5.5, Grok 4.3, Qwen3.7-Max) and...
Earth AI
A Blog post by Ai2 on Hugging Face
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
