psyops

35% in 3 days. also check the weekly, reclaimed a very significant level at 200 tao men.

You guys need to realize that 95+% of these AI workflow articles are written by the EXACT same type of person who used to waste all their time setting up VIM keybinds The primary goal of this person is not and will not ever be "productivity" Their goal is, in fact, the opposite

Two conversations this weekend make me think that there's a vibe shift afoot in Silicon Valley around what one should work on and what is worthwhile. Culturally, it feels like the moment is ripe for new frameworks: • Davos expert morality is stale and discredited. • It's also apparent that the "just be super based" Counter-Enlightenment is not really an answer. (Yes, woke went too far, but simply inverting it doesn't work.) • EA is no longer the automatic default for smart people. • There's increasing skepticism of slot and slop machine dynamics. Overall, "what is worthy and valuable?" feels like it's becoming more central.

this is binance's $BIFI daily chart lol christmas crimes up in this ho

NEW: BITCOIN SLIPPED IN ASIA TRADING HOURS WHILE OTHER RISK ASSETS GAINED AFTER THE US FEDERAL RESERVE CUT INTEREST RATES AND EXPRESSED OPTIMISM ABOUT THE ECONOMY SOURCE: bloomberg.com/news/articles/…

@0xkyle__ add me bro

DiLoCoX: A Low-Communication Large-Scale Training Framework for Decentralized Cluster Afaik, the largest public distributed training done, at 107B parameters with only a 1Gbps bandwidth! * Based on DiLoCo (FedOpt with inner AdamW, outer Nesterov) * Add also a pipeline, independent for each DiLoCo replica * Overlap outer gradient for a whole outer step, itself made of (fake numbers) 500 inner steps of roughly 1s each? so you have 500s to communicate 107B parameters over a 1Gbps bandwidth * however, that would take ~1680s to communicate, so they also do int4 quantization (1680s->420s) and low-rank (420s->210s) * And they also adaptively decrease the compression rank through training, but conversely sync more often to be ~iso-com Their experiments at 1B and 107B on Qwen show DiLoCoX to be almost as good as the *much more* costly Data-Parallel that communicate 1000x more. One caveat is that they do 4000 steps, but don't mention the token budget -- so i suspect it's quite undertrained.

@connorking @novora_ congrats connor

Extremely excited to announce the launch of my next venture: Novora. @Novora_ is a crypto-native proprietary investment firm. We invest across public and private markets and actively support ambitious teams building at the frontier. Novora’s mission is to help visionary entrepreneurs build transformative companies and protocols disrupting open capital markets and technology. At Novora, we don’t just fund the frontier — we forge it alongside our partners. For those that know me, crypto has been a deep passion of mine since day one. I entered the space in 2016 and went full-time in 2017. I launched my first crypto startup while in college. It ultimately failed, but it gave me a chip on my shoulder that I’ve carried ever since — one that fueled the next eight years of relentless work, learning, and growth. Fresh out of college, I took a leap and joined a then-tiny startup called MoonPay — fewer than 10 people at the time, completely bootstrapped. I was the first growth hire, focused on building out the growth and GTM function. In just over a year, we grew the team to 70+, laid the groundwork for a global payments business, and built one of the most transformative products in crypto. MoonPay would go on to raise one of the largest funding rounds in industry history, expand to 180+ countries, and onboard 30 million users. Over time, I realized that while building at MoonPay was incredibly rewarding, my heart was still tethered to the capital markets side of the industry. That led me to Arca, a then-early-stage asset management firm in LA. I wore a lot of hats: BD, IR, liquid investing. I helped personally raise over $35M from institutional investors, launched new initiatives, and deepened my understanding of asset management and capital formation. Being at Arca during the first wave of institutional adoption was an invaluable experience. In 2022, I had the opportunity to join Social Capital and work alongside Chamath to build out and run their first liquid crypto fund. At the time, they had made a few venture bets in the space but hadn’t yet entered liquid markets — so I came on to lead that effort from scratch. It was a unique experience, and I was fortunate to learn from several sharp traditional long/short portfolio managers. Although I was focused solely on crypto, many of the frameworks, disciplines, and mental models I absorbed carried over directly. At the end of the day, markets are markets — no matter the asset class. Unfortunately, during the 2022 drawdown, the firm decided to cut exposure to the asset class, however, the experience was deeply rewarding and left a lasting impact on how I think and invest. If you were involved in the industry in 2022, you know how brutal it was. At one point, it was the lowest low I had ever experienced personally. Life punches you hard in the gut sometimes. But during those moments, you realize how cyclical life can be — and you have to dig deep and decide whether to stay down or rise stronger. I took some personal time off at the end of that year to reflect on what I wanted to do next and how I wanted to spend my time. I felt a strong urge to return to building and operating — to contribute something productive to the space. That journey brought me to Irys, a small but mighty team building powerful infrastructure for the onchain world. When I joined, we were a 7-person team. No one really knew who we were. We didn’t have much of a footprint — but we had built something incredibly valuable. Over the next 18 months, we raised additional capital, generated 7-figures in annual revenue, scaled the team, and found our stride – driving impact at a global scale. That chapter was truly special. It brought me to NYC, allowed me to explore the world, and gave me lifelong friendships. In October of last year, I stepped away from the company to begin the next chapter — to pursue my own goals. I knew I wanted to build something of my own. After a lot of reflection and conversations with close friends and mentors, the vision started to crystallize. I wanted to create something that allowed me to pursue my true passions: investing and operating. I’ve long been inspired by investment firms in traditional markets that take an activist approach — operating alongside the companies they invest in and driving real outcomes. That philosophy stuck with me. Novora was born from a simple but powerful idea: That an investment firm in crypto shouldn’t just allocate capital, but should build alongside the teams it backs. That we can bring a principal mindset to early-stage partnerships. That activism, strategic support, and hands-on collaboration belong in this space — especially in public markets. And that it’s not only possible, but necessary, to blend investing and operating in a way that drives meaningful, long-term impact. Earlier this year, I came across a blog post by Graham Weaver titled “How to Live an Asymmetric Life”. Weaver highlights how principles from investing can be applied to personal growth, and most importantly, he emphasizes seeking opportunities where potential gains significantly outweigh the potential loss. In the blog, there are four key principles he outlines: Do Hard Things Do Your Thing Do It for Decades Write Your Story These words stuck with me. Because when you pursue meaningful work, commit to the long term, and take a bet on yourself — you unlock the chance for deeply asymmetric outcomes. Starting a company might be one of the most asymmetric things you can do in life. At a time when crypto feels directionless — when leadership is often missing, or worse, driven by ego over integrity — I believe there’s a real need for people to step up with purpose. I’m not perfect, but I aspire to lead with character, consistency, and class. To operate with integrity. To show up for others. And to lead by example — not just in what I say, but in how I build, invest, and treat people along the way. I’m incredibly grateful for every chapter that’s led to this one. Each role, each relationship, each risk — they’ve all equipped me to take this leap. Thank you to my family, friends, and mentors who’ve supported me along the way. I’ll never forget it. The best stories are the ones still being written. This is only the beginning. More life.

We've reached a major milestone in fully decentralized training: for the first time, we've demonstrated that a large language model can be split and trained across consumer devices connected over the internet - with no loss in speed or performance.

We plan to launch a few markets on @HyperliquidX once HIP-3 is live with help of some friends The CEX listing process is deeply broken and we must find new pathways to liquidity that are fair to all participants Permisionless tech enables us to opt out of broken systems

@kelxyz_ @jawa2690 I think having the YC tag gave them more legitimacy vs competition

@jawa2690 i thought many had the exact same referral schema tho just worked again?

why did axiom succeed so hard

Yesterday, Bloomberg wrote an article on Hyperliquid. Today, a Bloomberg podcast guest gives his thoughts on risk assets and gave HL a shoutout: - Risk asset allocation should include digital assets like Bitcoin and Hyperliquid - Most crypto assets are not revenue generating assets, but Hyperliquid is - If HYPE was a publicly traded stock, it would be ~4x the current market cap - Crypto could give birth to the next generation of banks and lending (sounds a lot like Jeff's vision!) Hyperliquid

AlphaEvolve, our new Gemini-powered coding agent, can help engineers + researchers discover new algorithms and optimizations for open math + computer science problems.  We’ve used it to improve the efficiency of our data centers (recovering 0.7% of our fleet-wide compute resources on average). We’re also using it in chip design and to speed up Gemini’s training, the very models underpinning AlphaEvolve itself —  an exciting flywheel of progress!

@alpinestar17 unreal team

Barcelona are just too good for anyone at the moment

@kelxyz_ If you have gemini roo then it’s integrated in google docs most for writing but ig that’s not what you’re looking for

@0xpsyops I didn’t like notebook LM. I find it is too unbalanced towards the sources I provide Proper system should be able to both draw from user inputs and proactively draw from outside sources, should augment my ability to manage my research, much more

is there an llm native google docs kinda thing out there or

@Saint_Pump well played

There is it. The trump put boys. Gg

@kelxyz_ @huntersolaire_ Agreed, unreal bend

@huntersolaire_ the second one was like messi vs liverpool level imo need ancelotti on the mic after the game "We just couldnt defend the freekick"

nah that is just a madness

@Ritesh_Trades Bayern backline is gone, inter winning this

So all of you betting on bayern ?

Today is the start of a new era of natively multimodal AI innovation. Today, we’re introducing the first Llama 4 models: Llama 4 Scout and Llama 4 Maverick — our most advanced models yet and the best in their class for multimodality. Llama 4 Scout • 17B-active-parameter model with 16 experts. • Industry-leading context window of 10M tokens. • Outperforms Gemma 3, Gemini 2.0 Flash-Lite and Mistral 3.1 across a broad range of widely accepted benchmarks. Llama 4 Maverick • 17B-active-parameter model with 128 experts. • Best-in-class image grounding with the ability to align user prompts with relevant visual concepts and anchor model responses to regions in the image. • Outperforms GPT-4o and Gemini 2.0 Flash across a broad range of widely accepted benchmarks. • Achieves comparable results to DeepSeek v3 on reasoning and coding — at half the active parameters. • Unparalleled performance-to-cost ratio with a chat version scoring ELO of 1417 on LMArena. These models are our best yet thanks to distillation from Llama 4 Behemoth, our most powerful model yet. Llama 4 Behemoth is still in training and is currently seeing results that outperform GPT-4.5, Claude Sonnet 3.7, and Gemini 2.0 Pro on STEM-focused benchmarks. We’re excited to share more details about it even while it’s still in flight. Read more about the first Llama 4 models, including training and benchmarks ➡️ go.fb.me/gmjohs Download Llama 4 ➡️ go.fb.me/bwwhe9