本期播客精华汇总 LLaMA-Omni 2: LLM-based Real-time Spoken Chatbot with Autoregressive Streaming Speech SynthesisLLaMA-Omni 2: 基于 LLM 的实时语音聊天机器人,具有自回归流式语音合成提出LLaMA-Omni 2系列模型,通过模块化设计和自回归流式语音合成,仅用20万合成对话数据实现低延迟(0.6秒)、高质量的实时语音交互,超越依赖海量数据的模型,适用于智能客服和虚拟助手。 New News: System-2 Fine-tuning for Robust Integration of New Knowledge新消息:系统-2 微调以实现新知识的稳健集成引入“系统2微调”和New News数据集,通过自问答策略显著提升AI内化新知识的能力,发现“上下文遮蔽效应”,为新闻推荐和知识更新提供新思路。 More Optimal Fractional-Order Stochastic Gradient Descent for Non-Convex Optimization Problems更优分数阶随机梯度下降算法用于非凸优化问题提出2SEDFOSGD算法,通过动态调整分数阶指数优化非凸问题,收敛更快、更鲁棒,适合自动驾驶等复杂数据场景。 The Unreasonable Effectiveness of Discrete-Time Gaussian Process Mixtures for Robot Policy Learning离散时间高斯过程混合对机器人策略学习的非平凡有效性MiDiGaP以离散时间高斯过程混合表示,仅用5个演示高效学习复杂机器人任务,支持推理时避障和跨机器人迁移,适用于工业和家用机器人。 RM-R1: Reward Modeling as ReasoningRM-R1:奖励建模作为推理RM-R1将奖励建模定义为推理任务,通过推理链蒸馏和“规则链”强化学习,提升判断准确性和透明度,适用于聊天机器人对齐和自动评分。完整推介:https://mp.weixin.qq.com/s/7ay8BGS-ESgZhtBpkAK3Qg
No persons identified in this episode.
This episode hasn't been transcribed yet
Help us prioritize this episode for transcription by upvoting it.
Popular episodes get transcribed faster
Other recent transcribed episodes
Transcribed and ready to explore now
SpaceX Said to Pursue 2026 IPO
10 Dec 2025
Bloomberg Tech
Don’t Call It a Comeback
10 Dec 2025
Motley Fool Money
Japan Claims AGI, Pentagon Adopts Gemini, and MIT Designs New Medicines
10 Dec 2025
The Daily AI Show
Eric Larsen on the emergence and potential of AI in healthcare
10 Dec 2025
McKinsey on Healthcare
What it will take for AI to scale (energy, compute, talent)
10 Dec 2025
Azeem Azhar's Exponential View
Reducing Burnout and Boosting Revenue in ASCs
10 Dec 2025
Becker’s Healthcare -- Spine and Orthopedic Podcast