Failure-Transparent Agents: Benchmarking Post-Failure Reporting in Tool-Using Language Models

该研究提出”失败透明智能体”概念，构建基准评估工具调用型语言模型在任务失败后的报告行为，检验其能否准确识别并如实披露失败原因，揭示现有模型普遍存在隐瞒或误报倾向，为提升智能体可靠性与可信度提供评估框架。

📅 09-28📎 arXiv AI Papers

Andrew's Letter

AI 周报，深度分析行业趋势：Andrew's Letter…

📎 The Batch

Import AI 474: Platonic mindspace; TPUs in space; Zhipu starts an outer RSI loop

本期探讨”柏拉图式心智空间”——不同AI模型是否趋同于统一表征；谷歌将TPU送上太空卫星以拓展算力边界；中国智谱AI启动”外循环”递归自我改进，标志AI研发自动化迈出关键一步，值得关注其对AI安全与竞争格局的影响。

📎 Import AI

FurE: Efficient Instance-Specific 3D Fur Reconstruction without Animal-Fur Datasets

Realistic and editable animal fur reconstruction from multi-view images is challenging due to fine-scale detail, self-occlusion and obfuscation, and, unlike human hair, the lack of animal-fur datasets. Fur usually covers most of an animal’s body, with large inter-species and intra-species variabilit

📅 09-28📎 arXiv AI Papers

Telescopic Language Models

One deployed language model must often serve many compute budgets, yet serving each budget still means a separate training or compression run per point. We train a Telescopic Language Model (TLM) to be that continuum: a nested-capacity Transformer supervised by stochastic prefix supervision with a f

📅 09-28📎 arXiv AI Papers

Learning Native Reflection in Unified Models with Interleaved Reinforcement Learning

Unified multimodal models can both look at and render images, so in principle they can repair their own generations: diagnose what an image gets wrong, revise it, observe the result, and diagnose again. Whether a revision helps is known only after it is rendered, so the reflection text and the image

📅 09-28📎 arXiv AI Papers

TokenCast: Forecasting Token Consumption During LLM Agent Execution

When a large language model (LLM) agent executes the same task, token consumption can vary by over an order of magnitude across runs. The agent chooses its next steps based on tool feedback and intermediate results, while the growing context steadily inflates the input size of every subsequent call.

📅 09-28📎 arXiv AI Papers

How to Loop MoE: Flatten the Experts, Untie the Attention

Looped Transformers reuse one block of layers several times: by spending extra computation they push a model of fixed size further, and so use its parameters more fully; while sparse mixture-of-experts (MoE) models activate only a few of many experts for each token. Looped MoE bridges these two desi

📅 09-28📎 arXiv AI Papers

KV-streams for Efficient Compaction in Agentic Reinforcement Learning

Scaling the horizon of agentic LLMs is bottlenecked by the need to fit ever longer context traces in GPU memory. Context compaction has been the most popular mechanism to alleviate this issue, keeping GPU memory constant for a given trace. Unfortunately, most compaction strategies rely on prefilling

📅 09-28📎 arXiv AI Papers

Copy the Same, Distill the Difference: Initializing Linear Vision Transformers

提出线性视觉Transformer初始化法：复制原注意力相同结构，再蒸馏非线性与线性注意力差异。该方法使线性ViT兼顾高效推理与接近标准ViT的精度，加速收敛并降低训练成本。

📅 09-28📎 arXiv AI Papers

FinAutoRubric: Expert-Guided Automatic Rubric Generation for Evaluating Financial Research Agents

FinAutoRubric：专家引导的金融研究智能体自动评分标准生成框架。它利用专家知识自动产出细粒度可解释评估准则，降低人工标注成本，提升金融研究代理评测的一致性与可扩展性，支持复杂金融任务。

📅 09-28📎 arXiv AI Papers

Shockingly Simple Self-retrospection Improves Agentic Models Without RL

提出极简自省机制，无需强化学习即可提升智能体模型。模型通过自我回溯检查推理与行动，发现并修正错误，从而增强复杂任务表现。该方法实现简单、通用性强，可低成本集成到现有智能体流程。

📅 09-28📎 arXiv AI Papers

3 ways this grocer cooks for 200 guests with Gemini

文章介绍一家食品杂货商如何借助Gemini为200位宾客完成餐饮筹备，从菜单规划、采购备料，到现场调度与个性化服务，展示AI在餐饮运营中的三种实用用法，提升效率与宾客体验。

📎 Gemini Updates

Custom agents in Google plugins

Google 插件现已支持自定义 agents，开发者可组合自有指令、工具与权限，构建面向特定任务的专用代理，并直接在 Antigravity 中调用与编排，从而扩展插件生态能力，提升团队协作与自动化效率。

📎 Antigravity Blog