TypeSafe launched Jev, a decision model returning labels and scores, prompting the release of at least 20 alternatives from various developers within four…
1 editorial report · 2 verified social mentions. The most authoritative report leads while later evidence completes the story.
Inspired by TypeSafe's Jev (a non-generative model that only outputs bounded decisions — choices, probabilities, calibrated confidence), I built two small decision models end-to-end on identical budgets and tasks to answer one question: which substrate makes the better Jev? Repo: https://github.com/stube123890-hue/jev-decision-lab Setup (V4, the main result) : - ~15.09M params each, budget-matched to ±0.02% - jev-v8a: linear recurrent backbone (my Gated Segmented State Space architecture from linear-attention-lab) + decision head - jev-transformer: standard attention backbone + identical head - New task: long-context retrieval-decision — given a 2k–8k character passage, decide which of 4 sentences appears in it (2000 train / 500 val / 500 test) - Trained on a Colab T4, everything reproduci
Open mentionThe release of decision-making AI models like TypeSafe AI's Jev and OpenAI's Decisions API is shifting security automation workflows. Instead of text generation, these models output probabilities to categorize alerts. This cost-effective approach allows teams to process 100% of alerts, addressing the 1% of real security incidents typically found in low-severity noise. However, performance remains depend… https:// en.hacks.gr/techniti-noimosyni -allazei-ti-diacheirisi-eidopoiiseon-asfaleias-stis-etaireies/ # ArtificialIntelligence # SecurityAutomation # OpenAI # TypeSafeAI
Open mention