Recent Posts

A team of Gemini 4 Argon agents beats Minecraft

Gemini 4 Argon 智能体集群通关《我的世界》：长程自主软件工程的新里程碑

10/16/2026

arxivaics.ai

Designing Transformers for Continuous Diffusion Language Models

为连续扩散大模型重塑 Transformer：LangDiT 架构解析

10/16/2026

arxivaics.ai

Rethinking Routing Exploration in Modern Mixture-of-Experts

反思现代 MoE 中的路由探索困境：MirrorMoE 如何化解共享专家与细粒度分片的内耗

10/16/2026

arxivaics.ai

Latent Compaction with Test-Time Training

测试时训练赋能潜层压缩：KV Cache 的 32 倍神经无损压缩

10/15/2026

arxivaics.cl

Scaling Discovery Requires Forgetting

扩展自主科学发现必须学会遗忘：打破智能体历史锚定效应

10/15/2026

arxivaics.ai

Your Transformer Is Secretly Elastic

你的 Transformer 原来是弹性的：无需重训的任意算力动态适配

10/15/2026

arxivaics.ai

DoubtLess: Training-Free Contrastive Decoding for Efficient Reasoning

DoubtLess：无需训练的对比解码如何为大模型推理瘦身

10/14/2026

arxivaics.ai

Two Clocks: Why Larger Rollout Batches Learn Slower but Fail on Schedule in Reinforcement Learning with Verifiable Rewards

双时钟假说：为什么大 Rollout Batch 在 RLVR 中学得更慢却按时崩盘

10/14/2026

arxivaics.ai

AI Models Can Provably Hide Arbitrary Capabilities

模型能够可证明地隐藏任意能力：数字保险箱击穿黑盒红队审计

10/12/2026

arxivaics.lg

Images Talk Back: Guiding MMDiTs by Suppressing Image-to-Text Feedback

图像会回嘴：MMDiT 如何通过抑制图到文反馈重拾提示词忠实度

10/12/2026

arxivaics.cv

RSI-Anything: Unlocking Recursive Self-Improvement with World Models

RSI-Anything：世界模型驱动的跨领域递归自进化

10/12/2026

arxivaics.cl

Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception

人赃俱获：探针如何捕获大模型的潜伏破坏与心口不一

10/9/2026

arxivaics.lg