ScrutAI security (@xsser_w) 's Twitter Profile
ScrutAI security

@xsser_w

Cyber security | AI pentesting | LLM safety && LLM security | web application security | mainly focus on LLM pentesting, If you're interested in, PM me

ID: 100656970

calendar_today31-12-2009 02:30:45

3,3K Tweet

380 Followers

1,1K Following

ScrutAI security (@xsser_w) 's Twitter Profile Photo

我觉得事物演化过程中一个很重要的指标是泛化能力,也就是适用于大部分情况才可以在生态位里做的很好才能保持生存,所以大模型产品也是,第一个指标就是泛化能力,任何只能在单一场景下生存的都不是好产品

ScrutAI security (@xsser_w) 's Twitter Profile Photo

其实有很多东西可以做 抽象了一下发现世界的很多问题的scaling law在那么几个维度 所以构造一个抽象组合的llm架构的公司就可以实现一些很可怕的目的 gmx被黑 和我假设的一些思路有点像 并且也验证了我对未来的判断。 接下来就是实践下自己的推理是不是真的正确

ScrutAI security (@xsser_w) 's Twitter Profile Photo

Cherry 的人这么蠢的吗? Agent还能自主学习? 是没看过rl agent的知识吗? Deepseek r1论文看懂了没?真逗 能动性是模型本身训练出来的,和agent毛关系也没,能不能找个懂post trainning的问问,学习一下。真逗

Cherry 的人这么蠢的吗?
Agent还能自主学习?
是没看过rl agent的知识吗?
Deepseek r1论文看懂了没?真逗
能动性是模型本身训练出来的,和agent毛关系也没,能不能找个懂post trainning的问问,学习一下。真逗
ScrutAI security (@xsser_w) 's Twitter Profile Photo

之前有很多傻逼程序员,觉得自己不会被替代。留给他们嘴炮的时间不多了,抓紧时间输出,看他们啥时候被迫脱掉长衫

Thinking Machines (@thinkymachines) 's Twitter Profile Photo

LoRA makes fine-tuning more accessible, but it's unclear how it compares to full fine-tuning. We find that the performance often matches closely---more often than you might expect. In our latest Connectionism post, we share our experimental results and recommendations for LoRA.

LoRA makes fine-tuning more accessible, but it's unclear how it compares to full fine-tuning. We find that the performance often matches closely---more often than you might expect. In our latest Connectionism post, we share our experimental results and recommendations for LoRA.