News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
arxiv arxiv-ai | 11d | Beyond the Vote: When Do LLMs Add Value to Retrieval-Based Tabular Classification?🔧 Consensus | Retrieval-based in-context learning (ICL) enables large language models (LLMs) to predict tabular labels from similar examples, but it… | ↗Not briefed |
arxiv arxiv-ai | 11d | Supervision-Adaptive Curriculum and Guarded Anti-Curriculum Learning for Dense Prediction🔧 Modal | Curriculum learning orders samples by difficulty, but difficulty alone does not establish supervision reliability for dense prediction… | ↗Not briefed |
arxiv arxiv-ai | 11d | Mitigating Query Shortcuts in Chain-of-Thought with Context-Aware Decoding | While Chain-of-Thought (CoT) prompting improves the reasoning performance of large language models (LLMs), recent studies have shown t… | ↗Not briefed |
arxiv arxiv-ai | 11d | BRIDGE: An Acceptance Gate to Block False Improvements in Model Updates, with a Retrospective Clinical Evaluation | Models are updated repeatedly, and an update that looks better on validation data but not on new data is a false improvement that can… | ↗Not briefed |
arxiv arxiv-ai | 11d | A unified approach for estimation of unnormalized models via Stein operators | In this paper, we develop a unified framework connecting proper scoring rules and Stein's method for the estimation of unnormalized mo… | ↗Not briefed |
bluesky bluesky-ai | 11d | 今日職場で聞いて面白かったのが、「LLMが出力する文章のLLMくささを消すことを「消臭」と呼ぶ」「最近のLLM界隈のトレンドの1つが消臭」っていうやつ。わざわざ「LLMくさいから使ってほしくないNGワードリスト」とかを作って渡すらしい。もはや書いたほうが早くね？ | 今日職場で聞いて面白かったのが、「LLMが出力する文章のLLMくささを消すことを「消臭」と呼ぶ」「最近のLLM界隈のトレンドの1つが消臭」っていうやつ。わざわざ「LLMくさいから使ってほしくないNGワードリスト」とかを作って渡すらしい。もはや書いたほうが早くね？ | ↗Not briefed |
hn hn-ai | 11d | Extension for Better VS Code Terminal Shortcuts | — | ↗Not briefed |
rss stratechery | 11d | Netflix Earnings, Business or Franchise, Disney’s Deal | Netflix has a solid business, which is exactly the problem: the company increasingly seems like a business and not a franchise, at lea… | ↗Not briefed |
arxiv arxiv-ai | 11d | Shortcut-Resistant CAM Distillation for Long-Tailed Recognition | Real-world datasets often follow a long-tailed distribution, making generalization to tail classes difficult. We revisit this problem… | ↗Not briefed |
arxiv arxiv-ai | 11d | Keeping Time vs. Understanding Time: Neither Alone Explains Temporal Reasoning in LLMs | Temporal reasoning is widely assumed to decompose into two modes: temporal arithmetic (date and time calculations), and temporal seman… | ↗Not briefed |
arxiv arxiv-ai | 11d | Asynchronous, Generation-Free ABC: Streaming AMIS Reweighting for Heterogeneous Simulator Workloads on HPC | Approximate Bayesian computation (ABC) enables Bayesian inference for models with intractable likelihoods by estimating parameter post… | ↗Not briefed |
arxiv arxiv-ai | 11d | Rethinking Agentic Model Orchestration for Time Series Forecasting | Time series foundation models (TSFMs) perform unevenly across forecasting tasks. Recent work uses agentic model orchestration (AMO) to… | ↗Not briefed |
rss techmeme | 11d | Anthropic and Citrini Research use an internal Claude model to autonomously generate protein minibinders against five scorpion neurotoxins, testing 211 designs (Anthropic) | Anthropic: Anthropic and Citrini Research use an internal Claude model to autonomously generate protein minibinders against five scorp… | ↗Not briefed |
arxiv arxiv-ai | 11d | Parallel Thinking for Large Language Model with Path Diversity | Parallel thinking has emerged as a promising paradigm for enhancing the reasoning capabilities of Large Language Models (LLMs) by expl… | ↗Not briefed |
hn hn-ai | 11d | Show HN: Linkit – AI bookmark manager, 100% free and open source | — | ↗Not briefed |
arxiv arxiv-ai | 11d | A Strong MoE Is All You Need for Process Supervision | Process reward models (PRMs) provide fine-grained supervision for multi-step reasoning, yet obtaining reliable process labels remains… | ↗Not briefed |
arxiv arxiv-ai | 11d | An Exact 93-Move Solution to a Nonlocal Crystal-Exchange Puzzle | We determine the minimum number of directional commands needed to solve alevel of Order of the Sinking Star, modeled as a crystal-exch… | ↗Not briefed |
arxiv arxiv-ai | 11d | RubricOpt: Self-Revision of Semantic Rubrics for Preference Optimization | A model's failures can reveal gaps in the behavioral criteria guiding its training. We introduce RubricOpt, a framework that lets a po… | ↗Not briefed |
bluesky bluesky-ai | 11d | Among Korean authors, the Hankyoreh also confirmed that the works by the late authors Park Wan-suh (“Who Ate Up All the Shinga?”), Ahn Jung… | Among Korean authors, the Hankyoreh also confirmed that the works by the late authors Park Wan-suh (“Who Ate Up All the Shinga?”), Ahn… | ↗Not briefed |
rss techmeme | 11d | Micron and the union of workers at its Taichung facility sign a multi-issue agreement, without giving details, resolving tensions at its largest Taiwan site (Reuters) | Reuters: Micron and the union of workers at its Taichung facility sign a multi-issue agreement, without giving details, resolving tens… | ↗Not briefed |
bluesky bluesky-ai | 11d | Glad more are realising this is a blunder! > By applying these tests outside their proper context [such as to LLM-based AI systems], the … | Glad more are realising this is a blunder! > By applying these tests outside their proper context [such as to LLM-based AI systems],… | ↗Not briefed |
bluesky bluesky-ai | 11d | Anthropic: we’re going to give away shares Also Anthropic: By the way, we lost $80 billion in six months, we keep losing control of our AI… | Anthropic: we’re going to give away shares Also Anthropic: By the way, we lost $80 billion in six months, we keep losing control of o… | ↗Not briefed |
hn hn-ai | 11d | GetDiogenes – I analyzed 200 tech job listings with Claude to spot ghost jobs | — | ↗Not briefed |
arxiv arxiv-ai | 11d | TenetBench: A Clause-Anchored Russian Corpus for Evaluating Language Models against Professional Codes of Ethics | Codes of professional ethics guide and restrict specialists, who increasingly work with assistants built on large language models. Suc… | ↗Not briefed |
bluesky bluesky-ai | 11d | 1A Geschäftmodell. | 1A Geschäftmodell. | ↗Not briefed |
arxiv arxiv-ai | 11d | Generative Modeling via Nonlinear Schrödinger Bridges and Branching Diffusions | We propose a new generative modeling framework grounded in forward-backward partial differential equations (PDEs). These equations ari… | ↗Not briefed |
arxiv arxiv-ai | 11d | Density-Level Backward Error for Low-NFE Generative Flows | Low-NFE generative ODE sampling is constrained not only by solver order, but also by how a fixed budget of network evaluations is dist… | ↗Not briefed |
arxiv arxiv-ai | 11d | Asymmetry Tunneling Beyond Adaptive Equivariance Relaxation | Equivariant neural networks encode prescribed symmetries as powerful inductive biases, yet real-world data may exhibit symmetry-breaki… | ↗Not briefed |
arxiv arxiv-ai | 11d | OmniSimulator: Aligning Small Language Models for Authentic Heterogeneous Behavior Modeling | Large language models (LLMs) have shown great potential as general-purpose user simulators for interactive systems. Prior studies have… | ↗Not briefed |
bluesky bluesky-ai | 11d | asks when Apple's call is scheduled, and prints the right line. To add it: connect your editor to Wirestack once, open the Voyage AI recipe… | asks when Apple's call is scheduled, and prints the right line. To add it: connect your editor to Wirestack once, open the Voyage AI r… | ↗Not briefed |
bluesky bluesky-ai | 11d | Interessant: Es gibt ja diese kuriose Situation, dass LLMs zT daran scheitern, Buchstaben in Worten zu zählen - zB wie oft der Buchstabe "R… | Interessant: Es gibt ja diese kuriose Situation, dass LLMs zT daran scheitern, Buchstaben in Worten zu zählen - zB wie oft der Buchsta… | ↗Not briefed |
bluesky bluesky-ai | 11d | It’s also why in the last week the valuation of Anthropic fell from $2Tn to $1.5tn, ehich is disastrous for every VC who invested in the co… | It’s also why in the last week the valuation of Anthropic fell from $2Tn to $1.5tn, ehich is disastrous for every VC who invested in t… | ↗Not briefed |
bluesky bluesky-ai | 11d | openAIとchatGPTがイコールになってない層がおるんかな かなり訴えられまくってるんすよね そういうところのサービス使うの普通に怖くねって思うんだが CEOも普通に倫理観無いし🔧 ChatGPT | openAIとchatGPTがイコールになってない層がおるんかな かなり訴えられまくってるんすよね そういうところのサービス使うの普通に怖くねって思うんだが CEOも普通に倫理観無いし | ↗Not briefed |
bluesky bluesky-ai | 11d | Las ficciones que exploran este tipo de personajes no tienen nada que ver con el producto que las empresas de los LLM y charcuterias genera… | Las ficciones que exploran este tipo de personajes no tienen nada que ver con el producto que las empresas de los LLM y charcuterias g… | ↗Not briefed |
arxiv arxiv-ai | 11d | PreCoMem: Predictive Cognitive Memory for Self-Evolving Long-Term Dialogue Agents | Long-horizon dialogue agents need to keep user memories stable enough for personalization while adapting when the user genuinely chang… | ↗Not briefed |
bluesky bluesky-ai | 11d | Unsaid in this article is what, if any, LLM was used to detect this vulnerability. | Unsaid in this article is what, if any, LLM was used to detect this vulnerability. | ↗Not briefed |
bluesky bluesky-ai | 11d | Welp, I may be chronically underemployed, but at least I made more than Anthropic in the first half of the year | Welp, I may be chronically underemployed, but at least I made more than Anthropic in the first half of the year | ↗Not briefed |
arxiv arxiv-ai | 11d | Harness Scaling for Generalizable Agentic Reinforcement Learning | Agent harnesses have become a central component of modern agentic systems, shaping how models use tools, manage context, and coordinat… | ↗Not briefed |
bluesky bluesky-ai | 11d | And now you know why Anthropic is very desperate for you to believe their whacko doomsday theories | And now you know why Anthropic is very desperate for you to believe their whacko doomsday theories | ↗Not briefed |
arxiv arxiv-ai | 11d | RegoMender: A Multi-agent Framework for Automated Rego Policy Repair | Policy-as-Code (PaC) has emerged as a foundational paradigm in cloud-native systems. However, repairing Rego policies remains challeng… | ↗Not briefed |
hn hn-ai | 11d | America's next big healthcare reform might be underway | — | ↗Not briefed |
bluesky bluesky-ai | 11d | Every time there's a new study on the negative impact of LLM tools on education, work, communications, and media, at least half the comment… | Every time there's a new study on the negative impact of LLM tools on education, work, communications, and media, at least half the co… | ↗Not briefed |
arxiv arxiv-ai | 11d | Omnilingual-MsASR: Benchmarking Multimodal, Multilingual, Long-Form Multi-Speaker ASR of Omni Large Language Models🔧 Modal | Omni-modal large language models (LLMs) have achieved remarkable progress in audio and video understanding. However, existing models s… | ↗Not briefed |
bluesky bluesky-ai | 11d | Why Anthropic can't afford to delay its IPO #Axios | Why Anthropic can't afford to delay its IPO #Axios | ↗Not briefed |
hn hn-ai | 11d | Agent Atoll: Visualise your Claude Code/Codex history on a 3D living archipelago | — | ↗Not briefed |
arxiv arxiv-ai | 11d | DynaWeave: Learning to Adapt through Dynamic Scenario Synthesis | Large language model (LLM) agents increasingly use tools to complete tasks in stateful environments. However, many training environmen… | ↗Not briefed |
bluesky bluesky-ai | 11d | OpenAI says its AI agents weakened the internal monitoring system www.reuters.com/legal/litiga... | OpenAI says its AI agents weakened the internal monitoring system www.reuters.com/legal/litiga... | ↗Not briefed |
arxiv arxiv-ai | 11d | How Far Can Foundation Models Be Distilled for Efficient Jet Tagging? | Jet tagging is one of the most important applications of machine learning at the Large Hadron Collider. Recent models build in more ph… | ↗Not briefed |
arxiv arxiv-ai | 11d | Events, Not Pixels: Where World Models Get the Return Wrong | World models are increasingly used for planning, where a planner ranks candidate actions by returns imagined by the model. However, th… | ↗Not briefed |
hn hn-ai | 11d | Ask HN: What should I do with 50k verified emails | I am working on my startup idea. I have collected about 50,000 verified names and email address of student and candidates looking for… | ↗Not briefed |