Caleb Fenton (@caleb_fenton) 's Twitter Profile
Caleb Fenton

@caleb_fenton

cyber security, machine learning, #Bitcoin, tech news, really mostly just Bitcoin memes

ID: 143071797

linkhttps://calebfenton.github.io/ calendar_today12-05-2010 14:18:00

1,1K Tweet

2,2K Followers

427 Following

Haider. (@slow_developer) 's Twitter Profile Photo

huge step toward self-improving LLMs this paper introduces a new way for llms to improve without external rewards the result: • matches RLHF on math • +60% improvement on code benchmarks • learns to reason step by step • scales from 1b to 14b • fast convergence in ~10 RL

huge step toward self-improving LLMs 

this paper introduces a new way for llms to improve without external rewards

the result:
• matches RLHF on math
• +60% improvement on code benchmarks
• learns to reason step by step
• scales from 1b to 14b
• fast convergence in ~10 RL
Caleb Fenton (@caleb_fenton) 's Twitter Profile Photo

Every day AI gets a little better. This means today is the worst AI will ever be. If you think it's hard to know what's going on *now*, I have terrible news for you. The next decade is going to be very interesting.

Caleb Fenton (@caleb_fenton) 's Twitter Profile Photo

Really enjoyed getting to nerd out about malware, reverse engineering, AI, and cyber warfare on the Complex Systems podcast with Patrick McKenzie: complexsystemspodcast.com/episodes/machi…

Vijay Boyapati (@real_vijay) 's Twitter Profile Photo

Why is Bitcoin's price stuck? There are billions in inflows from ETFs and treasury companies and the supply of newly mined Bitcoin is miniscule compared to these flows. What gives? One answer is that there's lot of paper Bitcoin flowing around suppressing the price. I do not

Caleb Fenton (@caleb_fenton) 's Twitter Profile Photo

TIL telling a model to "carefully review my instructions, your answer, and UNFUCK yourself" isn't good prompt engineering. But it does feel good.

Chubby♨️ (@kimmonismus) 's Twitter Profile Photo

SEAL: LLM That Writes Its Own Updates Solves 72.5% of ARC-AGI Tasks—Up from 0% This is a breakthrough that is rarely seen and could open up undreamt-of possibilities. In the following, I will go into more detail and summarize this breakthrough:

SEAL: LLM That Writes Its Own Updates Solves 72.5% of ARC-AGI Tasks—Up from 0%

This is a breakthrough that is rarely seen and could open up undreamt-of possibilities. In the following, I will go into more detail and summarize this breakthrough:
Bryan Johnson (@bryan_johnson) 's Twitter Profile Photo

Sleep deprivation will make you an animal and powerless to stop the bad instincts. Prioritize 7-8 hr each night. Build your life around it.

alphaXiv (@askalphaxiv) 's Twitter Profile Photo

Introducing your arXiv Research Agent A personal research assistant with access to arXiv + bioRxiv + medRxiv + Semantic Scholar. Upload drafts, conduct literature reviews, get insights across millions of papers MCP support coming soon 🚀

Delphos Labs (@delphoslabs) 's Twitter Profile Photo

Machine Learning Meets Malware. If cognition becomes an API call and malware can be reverse-engineered by an LLM, then what’s left of “zero trust”? Caleb Fenton joined Patrick McKenzie for a chat on AI, nation-states, and the new front in software security. 🎧complexsystemspodcast.com/episodes/machi…

Michael Luo (@azianmike) 's Twitter Profile Photo

I got a cease and desist from DocuSign for my free SaaS. A couple of months ago, I saw a tweet from Andrew Wilkinson: “I just found out how much we pay for DocuSign and my jaw dropped. What's the best alternative?” Me being naive, I thought “how hard could would it actually be to

I got a cease and desist from DocuSign for my free SaaS.

A couple of months ago, I saw a tweet from <a href="/awilkinson/">Andrew Wilkinson</a>: “I just found out how much we pay for DocuSign and my jaw dropped. What's the best alternative?”

Me being naive, I thought “how hard could would it actually be to