LIVE · refreshes every 20 min
updated Aug 23, 12:41 UTC
110010.art(ificial intelligence)
AI products, research, and launches — clustered and ranked, not just re-blogged.
110010.art / archive

Research

Aug 5
18d ago Adaptive sampling for automated post-disaster rapid damage assessment via level-set cost-aware Bayesian optimization arXiv cs.LG → story
18d ago Agentic coding in the wild: characterization of GitHub Copilot traces at production scale arXiv cs.AI → story
18d ago AgentStream evaluates self-evolving LLM agents under streaming tasks arXiv cs.AI → story
18d ago AI Agent Economics: can autonomous economic behavior emerge among AI agents under minimal external conditions? arXiv cs.AI → story
18d ago BAP-SQL: Budget-aware observation planning for agentic text-to-SQL arXiv cs.AI → story
18d ago Beyond the hivemind: escaping LLM homogeneity via meta-persona anchoring and sequential temperature scaling arXiv cs.AI → story
Aug 4
18d ago Machine-learning-guided pipeline enables on-demand solvent candidate generation and testing of electrolyte recipes; SEM images show sodium-metal deposit morphologies for three candidates MIT News (AI) → story
19d ago Benefits of medical AI assistance vary by user expertise, with non-experts trusting AI advice more than clinicians who recognize AI mistakes. MIT News (AI) → story
19d ago DiffusionGemma: an experimental open-weight language model uses discrete diffusion to generate text in parallel by refining blocks of 256 tokens; authors fine-tune mixture-of-e gates? Wait ensure accuracy. We must avoid inventing. The abstract ends with "fine-tuning the mixture-of-e" truncated. We can't infer. So craft headline: "DiffusionGemma introduces an experimental open-weight language model that uses discrete diffusion to generate text in parallel by refining blocks of 256 tokens." That stays within provided. Might be okay. Include arXiv 2608.00146? Not necessary. It's a headline style. Keep sentence case. Under ~200 chars. Let's produce: "DiffusionGemma introduces an experimental open-weight language model that uses discrete diffusion to generate text in parallel by refining blocks of 256 tokens." DiffusionGemma introduces an experimental open-weight language model that uses discrete diffusion to generate text in parallel by refining blocks of 256 tokens. arXiv cs.CL → story
19d ago DLLM-TTS: Block discrete diffusion language model for text-to-speech synthesis arXiv cs.CL → story
19d ago DSETA: a dual-stage continual learning framework for travel time prediction in dynamic traffic environments arXiv cs.LG → story
19d ago Dual-penalty evasion framework to fool white-box explainable AI auditors; arXiv:2608.00566v1 reports attacks on post-hoc explainers like LIME, SHAP, and Integrated Gradients arXiv cs.LG → story
19d ago Ensemble of unsupervised deep learning for clustering imbalanced tabular data arXiv cs.LG → story
19d ago Fairness auditing: quantify unavoidable post-audit manipulation under finite resources arXiv cs.LG → story
19d ago HP-JEPA: hierarchical partitioning for multi-resolution graph joint-embedding predictive learning arXiv cs.LG → story
19d ago Inference-time policy alignment for fair reinforcement learning arXiv cs.LG → story
19d ago Leak It: a probabilistic approach to training-data extraction from black-box language models arXiv cs.LG → story
19d ago Learning compositional meta-routing for agentic workflows; an executable benchmark arXiv cs.LG → story
19d ago LLantia: an LLM-based method for automated neural circuit inference and analysis arXiv cs.CL → story
19d ago LLM-OSDA: an optimal-stopping dynamic auction for native advertising in multi-turn LLM conversations arXiv cs.CL → story
19d ago MemoryForge: Synthesize lifelong memory for human-like LLM agents arXiv cs.CL → story
19d ago Modeling unknown nonlocal PDE systems via flow-map learning arXiv cs.LG → story
19d ago Neural operator learning for collision-aware trajectory planning of spacecraft swarms arXiv cs.LG → story
19d ago Obshazard-bench: benchmarking multimodal foundation models for real-time disaster intelligence from raw earth observation streams arXiv cs.CL → story
19d ago Physics-chemistry-informed neural network (PCINN) enables real-time SALD surface coverage prediction and reliable kinetics inversion arXiv cs.LG → story
19d ago Progressive^2: a teacher-student progressive co-evolving knowledge distillation method for substantial model compression arXiv cs.LG → story
19d ago Response magnitude as a dominant signal for held-out CRISPRi perturbation effect prediction arXiv cs.LG → story
19d ago Rethinking pretraining for specialized design data: evidence from the JONES-19 cultural design dataset arXiv cs.LG → story
19d ago Role steering of language models for social simulations arXiv cs.CL → story
19d ago RubricReviewer: converting direct critique into rubric-driven peer review to make reviews objective and comprehensive arXiv cs.CL → story
19d ago SeDeM: selective decompression of hidden-state memories for long-context question answering arXiv cs.CL → story
19d ago Similarity-aware machine unlearning arXiv cs.LG → story
19d ago Sixteen language models yield 1.69 distinct formulations on a psychotherapeutic case; ensembles offer multiple perspectives before a decision arXiv cs.CL → story
19d ago SLMs as multi-agent routers: a progressive SFT and reinforcement learning approach arXiv cs.CL → story
19d ago Stabilized best-of-K training for neural combinatorial optimization improves Leader Reward with a stabilized rank signal for POMO on TSP-100; achieves 7.7662 under 100-start, 8-augmentation greedy decoding arXiv cs.LG → story
19d ago Targeted low-rank adaptation via causal layer selection to bridge English-Arabic medical knowledge gap arXiv cs.CL → story
19d ago Textual descriptors predict early-stage startup exit using a 850-feature text-based framework on 7,419 startups over 20 years. arXiv cs.CL → story
19d ago Trustworthiness costs of domain adaptation in small language models; a cross-architecture empirical study arXiv cs.CL → story
19d ago Verifier-induced support reshaping in on-policy optimization shows verifiable rewards shaping trajectories and reducing successful behaviors for later objectives in RLVR arXiv cs.LG → story
19d ago What transfers from text to vision? Capability-Driven Multimodal Scaling Law and transfer dynamics for VLMs arXiv cs.CL → story