LIVE · refreshes every 20 min
updated Nov 14, 08:23 UTC
110010.art(ificial intelligence)
AI products, research, and launches — clustered and ranked, not just re-blogged.
110010.art / archive

Research

Oct 27
18d ago DriveMind: Mind-evolving belief tracking for closed-loop autonomous driving arXiv cs.AI → story
18d ago Each insufficient in its own way: calibrating language models across varying evidence states arXiv cs.CL → story
18d ago Effectiveness of curriculum learning depends on reward sparsity and competing optima arXiv cs.LG → story
18d ago Efficient omni-modal large language model inference via per-query modality selection arXiv cs.LG → story
18d ago Emergence and sparsification of a syntactic circuit in language models during training on subject-verb number agreement arXiv cs.CL → story
18d ago End-to-end differentiable diffusion conditioning for physics-informed optimization arXiv cs.LG → story
18d ago Energy-steered models are scalable foundation learners arXiv cs.CL → story
18d ago Entity-aware evaluation of OCR systems for historical document transcription shows that standard metrics miss meaningful information preservation arXiv cs.CL → story
18d ago EviGround-CT: sentence-to-volume grounding verifies LLM-generated chest CT reports by locating evidence for positive findings with a 3D grounding model arXiv cs.CL → story
18d ago Feature scaling beats architecture choice in offline respiratory screening for low-resource clinics arXiv cs.LG → story
18d ago Forecasting associations, not interventions: a preregistered causal audit of covariate-aware time series foundation models arXiv cs.LG → story
18d ago From Writing Advice to Manuscript Governance: A Constrained-Transformation Framework for Academic Revision arXiv cs.CL → story
18d ago Gaussian approximation for composition and subsampling of differentially private algorithms arXiv cs.LG → story
18d ago GEMS-RAG: graph-grounded engineering multimodal standards retrieval-augmented generation arXiv cs.AI → story
18d ago Harness self-evolution benefits vary by model capability; skill adaptations that help stronger models may offer little or negative value for weaker models. arXiv cs.LG → story
18d ago How pre-trained capacity affects causal learning after adaptation on Qwen 2.5-based CausalLLM and CausalGen; larger backbone size improves performance arXiv cs.LG → story
18d ago Individual fairness via local predictive consistency arXiv cs.LG → story
18d ago Inference-time prior adaptation for imbalanced data streams under concept drift using PAI-KNN arXiv cs.LG → story
18d ago Information-theoretic scaling laws for generative diffusion training arXiv cs.LG → story
18d ago Inverse modeling for laser pulse shape design in inertial confinement fusion arXiv cs.AI → story
18d ago IRON-Meme: a benchmark for structured understanding of identities and roles in online meme narratives arXiv cs.AI → story
18d ago Isharah-Selfie: Continuous sign language recognition dataset for one-handed signing arXiv cs.AI → story
18d ago Knowing whom to trust is not enough: the trust–agreement gap in repeated multi-party LLM negotiation arXiv cs.AI → story
18d ago Knowledge-guided LLM for intent-driven service orchestration in integrated terrestrial and non-terrestrial 6G networks arXiv cs.CL → story
18d ago Learning nonlinear waveform fusion across scales for ISAC image transmission and multi-target localization arXiv cs.AI → story
18d ago Learning structured sparsity for large models: bridging model compressibility and efficient inference arXiv cs.LG → story
18d ago Learning to Be Monitored: Causal bounds and intervention-aware post-training recipes for chain-of-thought monitorability arXiv cs.LG → story
18d ago LeWorldModel preserves local action geometry only loosely; the most sensitive action directions differ by median angles around 36–42 degrees across perturbation scales. arXiv cs.LG → story
18d ago Linear-Time context compression with state-space encoding arXiv cs.CL → story
18d ago Lower expected Fisher rank implies better PAC generalization arXiv cs.LG → story
18d ago LSTM-style recurrent transformer shares a key–value cache across its core, reducing parameters by 40.4% at width 1024 and lowering persistent cache sets from fifteen to four arXiv cs.LG → story
18d ago MARCA: a four-agent retrieval-augmented framework for Naranjo-based causality assessment in ICSRs; study compares agreement with human experts and label differences arXiv cs.CL → story
18d ago Merge-based vocabulary extension stalls at the pre-tokenizer for Nepali, delaying English-preserving repair during continued pretraining arXiv cs.CL → story
18d ago Mine or yours? Mind-Bridged language models take each other’s memories as their own arXiv cs.CL → story
18d ago MLIPDiff uses physics-informed guidance to steer diffusion models for structure-based drug design with interatomic potentials arXiv cs.LG → story
18d ago Model-agnostic predictability quantification of multivariable time series arXiv cs.LG → story
18d ago Multiple Importance Sampling for guided speculative inference applies to ensemble draft steps verified by a base model and a reward model arXiv cs.LG → story
18d ago Muon preconditioning improves with structured gradient noise, not isotropic noise, according to a study of Muon in large language model training. arXiv cs.LG → story
18d ago Nested sparsity for scalable pretraining and continual learning arXiv cs.LG → story
18d ago Neural-DISCO: source-conditioned counterfactual editing of neural population activity arXiv cs.LG → story

Older items: monthly archive →