Haolai(Anthony)Che

Excited to share CRUXpider (v0.1.0), an early open-source research asset discovery tool under the CRUX umbrella: crux-project.github.io. CRUXpider helps researchers go from just a paper title, topic, or research area to datasets, code, benchmarks, reading paths, and representative papers for AI4Science workflows. It currently integrates arXiv, OpenAlex, Semantic Scholar, Crossref, DataCite, OpenAIRE, and the GitHub API, and can be launched with a simple one-command local setup. This is an early release with much room for improvement, but we’re opening it early for feedback and collaboration. If this direction resonates, we’d love your thoughts, issues, stars, and contributions. Contact: hxc859@case.edu CC: ywx1650@case.edu, Dr. Yinghui Wu (@yinghuiwu4) @dbgroup_cwru @cwru GitHub: github.com/crux-project/C… #OpenSource #AI4Science #ResearchTools #MaterialsScience #MachineLearning

📎We've uploaded it to arXiv, enjoy! arxiv.org/abs/2603.15031

斯坦福大学的这篇论文值得了解👇🏻 他们构建了一个AI智能体框架，从零数据起步，没有人工标注，没有精心设计的任务，也没有任何演示，但它竟然超越了所有现有的自博弈方法。 这个框架名为Agent0：通过工具集成推理，从零数据中释放自我进化的智能体。 它所取得的成就令人难以置信。 你之前见过的所有“自我提升”的智能体都有一个致命的缺陷：它们只能生成比它们已经知道的稍微难一点的任务，所以它们会立即达到瓶颈。 Agent0打破了这个天花板。 关键在于： 他们从同一个基础LLM中生成两个智能体，并让他们竞争。 • 课程智能体 - 生成越来越难的任务 • 执行智能体 - 尝试使用推理+工具来解决这些任务 每当执行智能体变得更好时，课程智能体就会被迫提高难度。 每当任务变得更难时，执行智能体就会被迫进化。 这创造了一个闭环、自我强化的课程螺旋，而且这一切都是从头开始的，没有数据，没有人，什么都没有。 只是两个智能体互相推动，达到更高的智能水平。 然后他们添加了作弊码： 一个完整的Python工具解释器在循环中。 执行智能体学习通过代码来推理问题。 课程智能体学习创建需要使用工具的任务。 所以两个智能体都在不断升级。 结果呢？ → 数学推理能力提高+18% → 一般推理能力提高+24% → 击败R-Zero、SPIRAL、Absolute Zero，甚至使用外部专有API的框架 → 所有这些都来自零数据，只是自我进化的循环 他们甚至展示了难度曲线在迭代过程中上升：任务从基本的几何开始，最终达到约束满足、组合学、逻辑谜题和多步骤依赖工具的问题。 这是我们见过的最接近LLM中自主认知增长的东西。 Agent0不仅仅是“更好的RL”。 它是智能体引导自身智能的蓝图。 智能体时代已经解锁。

Google just dropped "Attention is all you need (V2)" This paper could solve AI's biggest problem: Catastrophic forgetting. When AI models learn something new, they tend to forget what they previously learned. Humans don't work this way, and now Google Research has a solution. Nested Learning. This is a new machine learning paradigm that treats models as a system of interconnected optimization problems running at different speeds - just like how our brain processes information. Here's why this matters: LLMs don't learn from experiences; they remain limited to what they learned during training. They can't learn or improve over time without losing previous knowledge. Nested Learning changes this by viewing the model's architecture and training algorithm as the same thing - just different "levels" of optimization. The paper introduces Hope, a proof-of-concept architecture that demonstrates this approach: ↳ Hope outperforms modern recurrent models on language modeling tasks ↳ It handles long-context memory better than state-of-the-art models ↳ It achieves this through "continuum memory systems" that update at different frequencies This is similar to how our brain manages short-term and long-term memory simultaneously. We might finally be closing the gap between AI and the human brain's ability to continually learn. I've shared link to the paper in the next tweet!

This paper from Tsinghua University and Shanghai Jiao Tong University received perfect scores (6, 6, 6, 6) at NeurIPS 2025! It aims to answer a key question: Does reinforcement learning really make large language models better reasoners? The authors study Reinforcement Learning with Verifiable Rewards (RLVR) and find that while it improves accuracy for small k, it doesn’t create new reasoning patterns—meaning the base model still determines the upper limit of reasoning ability. Across six RLVR variants, performance gains plateau, suggesting that current RL setups mainly refine reasoning rather than reinvent it. Interestingly, it’s distillation, not RL, that shows genuine signs of emergent reasoning. This research points to the next frontier for truly self-improving large language models. Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model? Paper: arxiv.org/abs/2504.13837 Page: limit-of-RLVR.github.io Our report: mp.weixin.qq.com/s/2-GDxs8j1QYh… 📬 #PapersAccepted by Jiqizhixin

@Shenxiao123971 🥕

最难吃的蔬菜投票

@FuSheng_0306 所以单纯的coding已经很cheap了，人类要做的是设计更好的算法，更优秀的理论，更好的架构，更高效的prompt.

腾讯内部人士透露：腾讯50%新增代码由AI辅助生成！你还觉得AI写代码不靠谱吗？

A paper about how to read a paper.

@yyds123888 那年我刚上大一，根本没有一百万，就算有认知也跟我没有关系

最近两天我操作挺傻逼的，亏了10个bnb。 今天无意中翻到一个快 10 年前的音频，是孙宇晨的。 听完我整个人都不好了。 他说：别买房、别买车。 拿 100 万，在 2016 年分成 5 份，买这几个东西——比特币、英伟达、特斯拉、腾讯，还有一个 LC。 拿到现在，6700 万。 我去，6700 万。英伟达和比特币翻了 150 倍。 就算里面有一个亏了 30%，那又怎样？ 在巨大的正确面前，这点小错误连个浪花都算不上。 最骚的是，他当时反反复复在念叨一件事：人工智能。 2016 年啊，我的天。 那时候我们都在干嘛？ 琢磨着首付，盘算着换车，觉得比特币是骗局，觉得特斯拉就是个电动玩具。 很多人骂他割韭菜，骂他戏精。 这些可能都是真的。 但你不得不服，人家在认知层面，真的把我们按在地上摩擦了。 我们赚的是辛苦钱，人家赚的是认知的钱。 这玩意儿，真比不了。

@mirrorzk 好文采。

我们该怎样告别呢？是从初遇的晚风说起,还是对往事的片刻回忆. 原来人生世事无常, 多的是别离 我们好像很讨厌等 等排队 等雨停 等风来 但我们又好像很喜欢等 等以后 等有钱 等有时间 我们一直都在等待 但我们不能一直只有等待 因为等不来春风得意 等不到功成名就 等不起岁月无情 这世界美好如斯，而我们仍然年轻. 大好光阴等我们去一步步完成夙愿，饱满灵魂，丰富阅历. 雪雨风烛，桑梓叶稀，顺从本辄，无人干涉. 于道各努力 千里自同风

Vibe coding is a wild experience. You enter a state of pure flow and emerge hours later with a masterpiece... that you don't fully understand, almost like a blackbox.

@mirrorzk 唐总这个上不上coinbase？

不要发币🥲 别人行我也行是一种错觉 你有好赌的爸爸吗？ 你有生病的妈妈吗？ 你有年幼的弟弟吗？ 你有家暴的老公吗？ 如果都没有，最好先算算自己命硬不硬 这里是地狱！

@micoolcho COOL! But that face is a little weird for sure

Selfie with the boss #IROS2025

So Humanoid Robot is the True Future.

@cryptopainter 可惜人只能活一次，你只能活成你自己，所以不需要活在别人的影子里，我虽然不是币圈的，但是“发偏财也是发财”

我可能是个失败者…给币圈丢脸了… 虽然一直在加密混，但这几年零零散散了解到了一些同学信息，当中有不少高中同学及大学同学混的都比我好… 有的是国内互联网大厂里面的小领导，还有的子承父业掌管了整个家族企业，最牛逼的一个同学，虽然不熟，但听说大学毕业后去干了淘宝店，年营业上亿的那种… 还有搞科研的，做理论物理的一个从小一起长大的朋友居然能够成功转行去了华尔街！？这是什么操作我到现在都没搞懂，理论物理和金融有什么直接的交集？ 另外就是两个高中同学，跑去美国读生物学博士，这么难以就业的职业居然也被人家俩跑成功了，最后留在美国还结了婚，成了真正意义上的学霸夫妇… 而我呢，除了炒币赚了点偏财以外，没干啥正经事业，未来如果有同学聚会，怕是也不敢参加了… 所以世界名校和一个亿选哪个？ 哎…

@BinsaeedRashid No he is not otherwise he'd say he is

JUST IN: Binance founder CZ believes Elon Musk is Satoshi Nakamoto.

@stats_feed He told you, He is Satoshi Nakamoto.

The anonymous Bitcoin creator Satoshi Nakamoto now holds over $121 billion worth of BTC — and no one on Earth knows who he actually is.

@cneuralnetwork I work in GNN explanation

Really want to connect with researchers on this app. If you are doing cool research at any lab, or academic research as an UG/PG/PhD, reply below - would love to connect! especially if you work in - LLMs - Machine Translation (low resource brownie points :D ) - Reasoning - Interpretability

Good resource, but only learn it when you need it. Don't frac your time and energy on random stuff.

@haydendevs Linux

what should I install first?