本期播客精华汇总 Phi-4-Mini-Reasoning: Exploring the Limits of Small Reasoning Language Models in MathPhi-4-Mini-Reasoning:探索小型数学推理语言模型的极限通过四阶段训练(大规模蒸馏、微调、偏好优化、强化学习),仅38亿参数的Phi-4-Mini-Reasoning在数学推理上超越70亿-80亿参数模型,揭示小模型需“量体裁衣”的训练策略,反直觉地发现朴素高质量数据可能有害。 ParamΔ for Direct Weight Mixing: Post-Train Large Language Model at Zero Cost直接权重混合的 ParamΔ:零成本训练后的大型语言模型ParamΔ通过简单权重差值加法,将后训练能力零成本迁移到新基座模型,性能达官方版的95%,为开源社区提供高效模型更新方案,揭示参数空间的代数结构潜力。 Model Connectomes: A Generational Approach to Data-Efficient Language Models模型连接组:一种面向数据高效的语言模型的方法受生物进化启发,提出“模型连接组”作为稀疏先验,仅用1亿词数据即可实现高性能语言学习,展现结构先验在数据效率和人脑对齐上的潜力。 Memorization and Knowledge Injection in Gated LLMs记忆与门控 LLMs 中的知识注入MEGa框架通过门控LoRA模块注入事件记忆,显著缓解灾难性遗忘,接近RAG性能,展示模块化记忆和内部回忆(iRAG)在持续学习中的前景。 AdaR1: From Long-CoT to Hybrid-CoT via Bi-Level Adaptive Reasoning OptimizationAdaR1:从长 CoT 到混合 CoT 通过双级自适应推理优化AdaR1通过融合长短CoT模型和双层偏好优化,实现自适应推理,推理长度减半而准确率仅微降,展现“因题施策”的高效推理潜力。完整推介:https://mp.weixin.qq.com/s/MyQN09CEBe59dbKcL7YEQg
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