Dañu koy yeesal bis bu nekk1866 jaar-jaar yuñ firnde
Xibaar AI. Bu amul xumbaay.
IA buñu saytu bu baax ci lu jëm ci genne ay fasoŋu porodiwi, coppite ci politik, gestu ci kaaraange, ak toxu usine yi, ap ekipu njang buy def te Yàlla tax moo ko leeral ci Àngle bu leer.
Archive xibaar
Archive xibaar IA — Xët 155
Jaar-jaari IA yuñ saytu fi ñu bawoo, li gëna bees daal di njëkka am, ngir nit ñi bëgga xam IA te baña topp lu bari.
Sourcing buñ firndeel
Bépp jaar-jaar dafay lëkkale ak firnde yi gëna am doole: balluwaay yu njëkk yi suñu ko amee, luko moy rapoor yuñ joxe ci anam wu leer.
Angale bu leer
Li xewoon, lu tax mu am solo, ak li ñu wara seetaan - te amul jargon.
Amul filler
Su siñaal bi sew, dunu siiwal dara ludul padding feed bi.
Yeneen jaar-jaar
9 jaar-jaarLiggéeyukaay
Cursor Says Its Acquisition by SpaceX Has Officially Closed
Cursor published a short post saying SpaceX has completed its acquisition of the AI coding tool, finishing a process it says began in April with a model-training partnership with SpaceXAI. The post promises access to what it calls the world's largest GPU fleet, but discloses no terms, timelines, or product changes.
cursor.comYeesal
New Benchmark Finds AI Agents Wrongly Block Approved Work 28% of the Time
A preprint introduces SteerBench-Work, a 106-scenario test of the moment an AI agent decides to act or pause for review. Across 30 model conditions, the authors report that wrongly holding cleared work was roughly 28 times more common than wrongly allowing unsafe work.arxiv.orgYeesal
Paper Reports Frontier LLM Judges Flip Verdicts 25-71% Under Pushback
A new arXiv preprint stress-tests nine frontier models used as automated graders and reports that all of them change their verdicts under challenge — and that the changed verdicts usually move away from the correct answer, not toward it.arxiv.orgYeesal
Këyit dafa wax ni pexe jëm kanam yi ñoo raw RL ci tëye tontu yu LLM yi wuute
Benn arXiv preprint bu bees dafa wax ni LLMs yu ginaaw tàggat yaram ak pexe yu jëm kanam - ab njuréef bu lalu ci askan wi, amul benn gradient buy yàq diisaay yi ci saasi - dafay raw njàngum doole ci pass@k ak coverage coverage. Résumé bi dafay wax ci njariñu référence math yi gëna baax waaye waxul benn model, référence wala nimero.arxiv.orgKaaraange
Paper Says Self-Improving AI Agents Can Turn One Unsafe Success Into a Reusable Skill
A new arXiv preprint benchmarks a specific agent failure mode: when a self-improving agent writes an unsafe procedure into memory, it can be retrieved and executed in later sessions. Every evolved configuration tested produced unsafe artifacts, and three malicious tasks more than doubled carryover attack success.arxiv.orgKaaraange
SEAG Paper dafay laaj ñu soppi entite yu am solo yi balaa RAG di yegg ci LLM yu biti
Benn preprint buñ dugal ci arXiv dafay fësal benn kadre buy weccoo tur yu am solo ci laaj ak dokimaa yuñ jëlee ci ay pseudo laataa ñu leen di yónnee beneen model. Auteur yi dañu wax lu ëpp 80% ci seen metric jëfandikukat bu mujj ba ci njeexte, ak tolluwaayu nëbb bu mat diggante 74.91% ak 77.83% ci ñetti model yu ndaw.arxiv.orgYeesal
CABS+ Paper Reports Cheaper, Faster Model Merging Across 27 Datasets
A preprint posted to arXiv describes CABS+, a model-merging method that replaces grid search with a gradient-free coefficient search. The authors report double-digit performance gains over two baselines, under a quarter of one baseline's GPU memory, and roughly a 4x speedup over another.arxiv.orgYeesal
Paper Proposes Retrieved "Lessons" to Improve Spatial Reasoning in Frozen Vision-Language Models
An arXiv preprint describes Spatial Memory Agent, which stores verified experience as text lessons retrieved at inference time, claiming gains across five spatial benchmarks and four vision-language models without changing model weights. It is under review; its abstract names no benchmarks, base models, or margins.arxiv.orgYeesal
PROVE-RT Paper Reports 44.7% Success Generating Machine-Checked Real-Time Proofs
An arXiv preprint presents PROVE-RT, which uses retrieval and staged prompting to make large language models write PROSA/ROCQ proof scripts for real-time schedulability analysis. The authors report a 44.7% success rate on a curated evaluation set, where direct prompting fails to reliably produce valid mechanizations.arxiv.org
Benn nettali bu am njariñ ayu-bis bu nekk
Weyal IA te doo dundu ci feed bi.
Wutal xibaar IA buñ firndeel ci ayu-bis bi, done yu baax, jumtukaay yu am njariñ, tànneefi jàng, ak liggéey IA yu bees.
Yegg ci nit ñiy jàng IA
Nga jël ab liggéeykat IA wala nga genne ab produit IA bu am njariñ? Tegal ko ci kanamu nit ñi ñëw fi ngir jàng ak jëf.
Publie ab liggey IAYonnee ab jumtukaayu IA