Planet AI Developer Filter
Lipid composition linked to stripe regularity in a lab model
A preprint links lipid composition to stripe regularity and pitch to stripe spacing in a laboratory liquid-crystal–water interface model.
Planet AI Developer Filter
Proofreading model links accuracy to the driven layer
A preprint models kinetic proofreading and finds that the error floor varies with drive position, while stronger drive may later raise modeled error.
Planet AI Developer Filter
Preprint sets error bounds for conductivity recovery from noisy scattered data
A preprint derives high-probability error bounds for recovering conductivity from noisy scattered measurements and tests sampling effects in simulations.
Planet AI Developer Filter
AI training-data filtering linked to stronger results after GRPO
A preprint reports higher pass@16 for TailSFT in most language-model tests and higher post-GRPO pass@1 in every matched comparison.
Planet AI Developer Filter
Quantum-Inspired Model Reports Lower Error Forecasting Cloud Workloads
A qubit-based cloud workload model reported lower benchmark prediction errors, but simulated tests did not establish benefits for live resource reservation.
Planet AI Developer Filter
Simulation links black-hole jet switching to magnetic flux
A preprint models black-hole jets that vary with magnetic flux and topology, using synthetic total and polarized emission for comparison with Sgr A⋆.
Planet AI Developer Filter
ReRAM Model Maps Memory States Across Two Switching Tests
A ReRAM model matched measured switching behavior and found corresponding equilibrium states in pulsed and quasi-static simulations.
Planet AI Developer Filter
OpenClaw Releases OpenClaw 2.0: Guided Model Setup, 575 ms Control UI Startup, and One Trust Boundary Per Gateway
OpenClaw 2.0 (v2026.8.1) lands 16,000+ pull requests, a rebuilt Control UI, SQLite session storage, and shared multiplayer sessions.
Planet AI Developer Filter
Turkish AI preprint reports retrieval gains with a training switch
A Turkish AI preprint reports stronger retrieval after CLM-to-MLM training, while single-seed tests leave the full-scale result unconfirmed.
Planet AI Developer Filter
Keystroke Data Sharpen Early Signals of Struggling CS1 Students
A preprint finds keystroke editing features improve early CS1 struggle prediction, with the strongest signal in the first submission segment.
Planet AI Developer Filter
Coupled-map model yields Gaussian fluctuations as systems grow
An arXiv preprint shows how mean-field coupled maps approach limiting laws and Gaussian fluctuations as particle numbers grow under strict assumptions.
Planet AI Developer Filter
EXAONE AI model leads on several tabular benchmarks
An arXiv preprint reports EXAONE Tabular led several regression tests and TabArena classification while delivering strong results at lower inference cost.