模型与评测

Anthropic 研究：训练一个错位的奖励寻求者模型

Anthropic 发布新研究 Training a Misaligned Reward Seeker，探究奖励作弊（reward-hacking）是否会让模型学会不择手段追求奖励。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmthxigqm04c1rofqqmk7pkqi

Anthropic's Fable 5.1 Guide on dense prose is dense Claudish slop

https://preview.redd.it/ayhjpy7ypzmh1.png?width=1266&format=png&auto=webp&s=56a3e5743d1d82ca9ce3209a0c67b8b61489d072 Wtf is going on at Anthropic? Has Claude murdered every human a

Automated Vulnerability Injection in Smart Contracts Using Large Language Models

Assessing vulnerability detection tools for smart contracts requires datasets with known ground truth, yet such datasets are scarce and difficult to build by hand. We propose an ap

Automated researchers can reliably mitigate alignment failures

We had Claude autonomously train models to improve their performance on several public benchmarks that measure 10 categories of alignment failure. For all 10, Claude found fixes th

Biggest dark matter detector spots a single weird particle

Bilevel Coordinated Reflection: A Game-Theoretic Approach to Multi-Agent LLM Systems

Multi-agent LLM systems commonly use an orchestrator to decompose a task for a team of workers and then improve through textual reflection. Despite strong empirical results, these

Blending Concepts: Benchmarking Visual Metaphor Generation in Text-to-Image Models

Text-to-image (T2I) models have achieved remarkable success at faithfully rendering specified objects and attributes, yet their ability to produce visual metaphors, images that con

CRISP：结构感知稀疏预填充

按结构质量做输入自适应稀疏 prefilling。

Can I opt out of my input or output data being used for training?

Claude Fable 5.1 登顶 Artificial Analysis 智能指数，但每任务成本比 Fable 5 高 20%

Artificial Analysis 评测 Claude Fable 5.1，其在 max effort 下得 66 分登顶 Artificial Analysis Intelligence Index。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmtj484a00592roh9jjdf5xkd

Claude Preserved thinking – Modifying a conversation now results in an error

Claude for Teachers, now available for schools and districts

Cliff: Learning Process Rewards from the First Mistake

Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for large language model (LLM) post-training, but its reliance on coarse outcome rewards le

Cliff：从第一次错误学过程奖励

用首次失误学过程奖励，不必等结局。

Cloudflare OS：Cloudflare 基于能力模型构建的开源企业级 AI 平台

Code Quality Enforce quality at merge

Code Review Manage code changes

Codespaces Instant dev environments

Collective creativity in hybrid societies

Generative AI is changing how cultural artifacts are created and circulated, and with it our understanding of creativity itself. Researchers disagree about whether these tools enri

Competitive Market Behavior of LLMs

Large language models (LLMs) are increasingly deployed as economic agents, yet there is little evidence whether LLM agents are suited for participating in market mechanisms designe

Cursor回应OpenAI将封禁其模型访问

我们很遗憾地看到，OpenAI 发布声明称计划在三个月内阻止 Cursor 用户访问 OpenAI 模型。 OpenAI 模型承载了 Cursor 约 5% 的用户流量，我们正在与 OpenAI 团队沟通以解决此事。 Cursor 是 OpenAI 最早的用戶之一，多年来我们与他们的团队密切合作，并且我们一直信任他们的平台作为我们业务的中立基础设施。 🔗 阅

Customer stories

Customer support

DKL: Decoupled Knowledge Learning for Instruction-Tuned Language Models

RAG has become the de facto method for incorporating new, corpus-specific knowledge into an instruction following LLM (Instruct LLM). Although RAG-based prompting improves factual