0/n After almost 3 years, it is with great pride, respect, and love to announce that February was my last month @binance. It's been a whirlwind of a ride, so what better way to share the memories, nostalgia, and future alpha I learned all in one tweet thread
here's the order i'd go into local ai, anon:
> 12gb: learn the stack on a compressed 27b, how a model gets served, how an agent runs overnight with features like hermes agent's /goal. you will see good results and see some light at 50 tok/s fresh.
> 24gb: this is where you meet the king, none other than the one and only 27b dense q4. get 2x 3060 if you want hands-on tensor parallel, or one 24gb card if you want to scale correctly and efficiently. at 24gb i prototype about 30% of my daily work, that's where owning your thinking starts.
> skip macs and the 5090 and the rtx 6000 pro for now, learn the stack small first and understand the difference. it's not a thing that happens overnight, it's a process and a lifestyle, you don't just start doing everything, get disappointed and then never realize how profound this is, that you can think and it stays on your metal.
> dgx spark: go gentle, and if you don't want to learn pcie lanes, power distribution and bifurcation, the dgx spark path is a clean cuda box, plug in and go. you can stack more and with the connectx cable it spits well. i have 2 dgx sparks so i can say for sure scaling with dgx spark is the cleanest. sure you don't get super fast bandwidth for dense models, but moes like qwen 3.8 flash next, ling flash and stepfun 3.7 dance on a single dgx spark and go zoom.
> but if you want to learn real infrastructure, from cable management to pcie lanes at x8 and x16, bifurcation, slimsas cables, backplanes, airflow and power distribution, that's what the dgx spark skips for you, and the price is bandwidth. the ultimate path is the rtx 6000 pro path if you dare touch fire, and the fire doesn't care about dense or moe, it just burns everything, which i only intend to.
rejoice my fren
AI agents install packages, run commands, touch wallets & ship code. Everyone is racing to make AI agents autonomous. I’m building the layer that can tell them “no.”
Tirith v0.4.0 adds:
• Python package firewall
• Web3 command guard
• Untrusted-project containment
• CI artifact-poisoning detection
• ThreatDB v2
• 19 integrations and more features
https://t.co/SJhVJjqX87
hi @cass_on_mars want to respect the time youve spent in being an active contributor to farcaster, but i want to call out this subtweet as not really useful or conducive to having direct conversations that bring the decentralized spaces forward, so ill speak the unspoken
@suji_yan suji, who you're trying to call out, has actively done many things to keep lens alive for the right reasons, and has been a beacon synonymous to social stuff within asia circles for longer than my entire career within crypto
suji has been an active angel and supporter of various social products ive gotten to use, and also a prolific supporterof many projects during their worst times, definitely not a passerby trying to just play a private equity game. as an angel investor in @impossible_
i of course have my own bias, but i speak for many users especially in asia who opened our minds towards decentralized social because of suji and team over these years.
calling anyone out over trying to help other platforns as just "failed social protocols", well, heres the reality of everything: you're posting about this on x.
And even zuck posts about meta launches on x.
so many users still sit on this platform today.
if we want decentralized social to still be a thing, itll take a full village and global voices to speak up for what they believe in and build the spaces we want.
and some of the bigotry ive seen in some farcaster chats talking about suji's personal choice of looks etc, sound exactly not like the type of environment to actually have constructive though around building better platforms for people.
@suji_yan suji has been someone thst ive always enjoyed brainstorming ideas with, in depths of bear market, in times of duress, in times of opportunity. which i cant say that for a lot of investors and a lot of my closest defi friends anymore, sadly
i also highlight that how acquisitions work is exactly that someone new is running the show. if you didnt like how lens was run in the past by aave, i think suji also is very open to hear how lens can be run better.
for some people in many countries around the world, and I can say, especially in Asia, uncensorable social is a life or death choice, and one worth fighting for for many, so this is saddest to see when I see one of the few people left who still have the balls to speak up for decentralization today, gets hammered with this type of extreme community backlash. i just wish that two OGs can actually have real dialogue on actually a lot of the same commonalities you all share, not just assumptions around background etc. on a person. Isn't that what having a neutral ground of a decentralized social platform is all about?
Tomorrow at 5pm PDT on Farcaster Spaces, I'm hosting a community call to discuss this, our ongoing plans and work on Hypersnap, and what our plans are around forming our proposal to run FC.
No idea if they'll accept our proposal, but I am one of the original core FC devs, and have been building Hypersnap with the core purpose of bringing FC to full decentralization and ability to run without a central benefactor sustaining it.
The last thing I want to see is FC end up in the hands of a private equity firm or a holding company of other failed social protocols (you know who I mean). For all those interested, we're happy to see you attend.
https://t.co/AiA6i3wh7I
@ns123abc I hope they get some pushback on this.
If I'm paying for your plan, I don't want invisible watermarks embedded in my content.
If this is really the direction they choose, we should seriously consider open source alternatives that don't impose this kind of control.
We have a surprise…
We’ve actually been working on Solana for months.
Today, we’re incredibly excited to finally say it:
Rarible is now live on @solana.
We're kicking things off with @Claynosaurz as the first featured collection on the marketplace.
This is just the beginning. Over the coming days and weeks, we'll continue onboarding more collections from across the Solana ecosystem.
Over the past few months, we’ve been heads down building, testing, and preparing for this launch. But technology was only one part of the journey.
We’ve spent weeks talking to NFT communities across Solana, diving deep into projects, understanding what makes each community unique, listening to feedback, and learning how we can build a marketplace that truly serves collectors and creators.
That feedback has already shaped many of the decisions we’ve made, and it will continue to shape everything we build from here.
If you’ve been paying attention, our Gacha Station launch on Solana was never just another feature. It was the first glimpse of what was coming.
Every project has its own story, identity, culture, and community. Our goal is not simply to list collections, but to work alongside the people behind them and create experiences that reflect what makes each project special.
What you see today is only the foundation. We’ll continue adding projects, features, marketplace improvements, content, and community activations as we work toward a much broader evolution of Rarible.
We’re building it step by step, alongside the Solana ecosystem, not just for it.
To everyone who shared feedback, challenged our ideas, tested our products, and believed in what we’re building: thank you. This launch belongs to all of you as much as it does to us.
We’re excited to write the next chapter of Rarible together with the Solana community.
Let’s build.
https://t.co/ilk4ShzR4p
WE ARE LIVE.
Introducing VARIBLE, the data and indexing platform built for the tokenized collectibles market.
Tokenized trading cards and physical collectibles are growing across platforms, blockchains and categories. But the data is fragmented, it's hard to see what's trading, where activity is happening, and how the market is actually performing.
VARIBLE brings it together in one place.
At launch, we index 140,000+ collectibles across four tokenized trading-card platforms (and growing), covering Pokémon, One Piece, Yu-Gi-Oh!, sports cards and more.
Discover:
- The biggest sales across the market
- Cards gaining trading momentum
- Leading categories and IPs
- Volume and market share by platform
- Marketplace vs. primary gacha activity
- Performance over 24h, 7 days and 30 days
- The market vs. Bitcoin, Ethereum, the S&P 500 and NASDAQ
- And the V-MKT Index: a single view of how the whole tokenized collectibles market performs over time.
This is not another marketplace. It's the data layer for an emerging onchain asset class.
Real cards. Real prices. Indexed.
https://t.co/AvH6cHhbn7
A @rarible project.
Its runninggg, LFGGG
New Deepseek-V4-flash-0731 @UnslothAI UD-IQ1_M is running on my 3090!!
Using all of my vram and ram here to generate those precious tokens!
Time to see how good it is.