NftPillQueen 🍭

FREE @NUBCAT

Assets added to the roadmap today: Mamo (MAMO) coinbase.com/blog/increasin…

No one is ready for Rei Core 0.3. ​ It will start to change the way we view the use of large language models altogether. ​ @ReiNetwork0x is a beast in the making. As many of you know, Rei is taking a novel approach when it comes to the use of LLMs and how we should interact with them. ​ Today, we simply assume LLMs can't be trusted and that we must verify each bit of data to confirm their output. ​ While we treat LLMs as 'human', it's just predicting which words need to be put after the first and providing an answer; there is no 'thinking'. ​ This is unacceptable if we want LLMs and AI to play a larger role in our lives, as it has started doing for so many of us. ​ We need to give LLMs both cognition and the ability to remember. ​ There is still a debate on how to approach these issues, but I believe the more we mimic biology, the better the output. ​ We should strive for an agentic system, specialized agents working together in harmony. ​ Rei shares this vision and built their bowtie, which changes how queries are approached compared to models like ChatGPT, for example. Rei is already showcasing its reasoning skills and has proven to be a strong competitor with their Core model. ​ Since its inception, Rei Core has proven to work as intended, being able to tackle many standard tests and closely compete with models 10-12 times larger than itself while scoring within the same range of ChatGPT. ​ We will need to see an updated benchmark test for Core 0.3 and see how much better it has gotten, but it's safe to say we'll see some major improvement, most notably in the reasoning tasks, which make up 3/5 of the test. Besides reasoning, memory is a core aspect of building a proper language model. Without it, models tend to hallucinate, and conversations become less reliable with every query you send. ​ In order for a model to think, it must be able to remember. ​ Core 0.3 will allow units to gain access to their own memory structure, also known as metacognition, which will also allow us to do the same. ​ This gives us real-time insights into which components of Core are working on your query, and means we can go far more in-depth to verify where an output came from. ​ We can view this using hypergraphs, making things more visually sensible, which is a great addition for us to better understand how Core operates and which components are more active when handling a specific query. ​ Core 0.3 is a huge improvement compared to the LLMs we've experienced so far, and it will show. There is still a long way to go, but Rei is making fast moves that allow them to compete with the strongest models we know on a far more efficient scale when it comes to cost and compute. ​ All the while, funds are taking a keen interest, and Core is being used for Ecliptica with a clear revenue model in place. ​ Name one project with a market cap of around $115M that has numerous advantages, continuously ships, and is already developing a real-world use case. ​ At some point, Rei isn't a beast in the making; it becomes one. ​ $REI remains one of my top conviction investments, and I see it as one of the easiest picks this cycle. ​ $1B is much closer than you think. ​ I hope you enjoyed this post 🫡 ​ These posts take a long time, so I'd appreciate it if you could like and retweet it.

My $REI journey: From $10M to over $135M by digging into fundamentals, tracking product-market fit, and following smart money using @nansen_ai ⸻ On June 23, I shared a post about $REI (see QRT) that ended up going viral with over 32k views. Honestly, I was a bit surprised. Especially because I don’t have a large following, and $REI isn’t exactly a mainstream name yet in the web3 trenches. There’s barely any (big) KOL coverage, and their main X account still sits at 12K followers. In that post, I shared my insights and findings on $REI and judging by the response, it seems people still appreciate real research and clear reasoning. So in this post, I want to walk you through how I managed my $REI position and the steps I took that made it become my third-largest portfolio position as of today, right after $TAO and $BTC. For the record, I’m not here to flex a 13x return (though I won’t complain). What really excites me is that I followed a process built through 8 years of experience, trial, and error (and many dollars lost) that actually worked. And I didn’t do it alone. I use a bunch of tools, but if I had to pick just one, it would be Nansen. Nansen is by far one of the best analytics platforms out there. It’s not cheap, but it’s worth every dollar. Without it, I wouldn’t have made this call as confidently or as early. If you’re curious to try it yourself, there’s a 10% discount link below in this thread. Let’s dive in 👇🧵