Tag
NandakishorM 开源了基于物理的类型化决策模型 Vega（800M 及 4B 参数版本），采用 Test-Time Training 引擎+适配器架构，支持 73k 上下文和图像输入，通过在隐藏层构建"结果山谷"并用带摩擦的小球物理模拟来判定输出意图，单次推理超越多项 JEV 基准。
The paper evaluates when zero-shot LLMs can replace fine-tuned NLU classifiers for intent detection in production, highlighting LLM advantages in out-of-scope detection, ASR robustness, and dynamic schemas, and offers a decision framework for practitioners.
The paper proposes a multi-cluster boundary learning method for out-of-scope intent detection using MiniLM embeddings, achieving state-of-the-art performance on CLINC150, StackOverflow, and Banking77 datasets.
A developer shares four key engineering decisions behind building a low-cost sarcastic Q&A AI side project, DumbQuestion.ai, covering persona consistency, self-awareness detection, freshness-sensitive routing, and async analytics, emphasizing cost-effective trade-offs and architectural choices.
This paper introduces Bot-Mod, a moderation framework that identifies malicious intent in multi-agent systems through multi-turn dialogue and Gibbs-based sampling, and presents a dataset from Moltbook for evaluation.