๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript) header
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript) profile picture

Project Management โ–ช๏ธŽAI Automation Top 1% in my craft...

Rome wasn't built in a day...
Joined July 2025
Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)Photo by ๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ (@Myke_Kript)
Managing crypto exposure doesn't have to mean restricting your portfolio to native tokens alone. โ€‹On @FBYTio, vault managers can execute strategies across both spot crypto assets and tokenized stocks directly on Solana.
2
1
19
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
REPORTING..๐Ÿ”ฅ
16
61
541
8,611
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
Weโ€™re triple trending โค๏ธโ€๐Ÿ”ฅ I see U N I T Y This is how we move as a family !!
WE ARE STILL TRENDING! Consistency looks good on this family โค๏ธโ€๐Ÿ”ฅ 22 days in a rowโ€ฆ
86
30
2
129
1,473
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
โ–‘โ–‘โ–‘โ–’โ–’โ–“โ–ˆ ษข ษช แด  แด‡ แด€ แดก แด€ ส แด› ษช แด แด‡ โ–ˆโ–“โ–’โ–’โ–‘โ–‘โ–‘ ๐ŸŽ ๊œฐ ส€ แด‡ แด‡ แด ษช ษด แด› ๊œฑ แด˜ แด แด› ๊œฑ ๐ŸŽ แดก แด‡ ' ส€ แด‡ ษข ษช แด  ษช ษด ษข แด€ แดก แด€ ส 8 ๊œฐ ส€ แด‡ แด‡ แด ษช ษด แด› ๊œฑ แด˜ แด แด› ๊œฑ ๊œฐ ส€ แด แด ๐Ÿพ ส™ ษช แด› ๊œฑ สœ แด แด แด… ! ๐ŸŽฎ ๐Ÿพ ส™ ษช แด› ส€ แด‡ แด› ส€ แด แด‡ x แด˜ แด‡ ส€ ษช แด‡ ษด แด„ แด‡ แด ษด ส€ สœ โžœ ๊œฐ แด สŸ สŸ แด แดก @8bitshood x @Rbornn1 โžœ สŸ ษช แด‹ แด‡ โค๏ธ โžœ ส€ แด‡ แด˜ แด ๊œฑ แด› ๐Ÿ” โžœ แด… ส€ แด แด˜ ส แด แดœ ส€ แดก แด€ สŸ สŸ แด‡ แด› ๐Ÿ‘‡ โณ ๐Ÿท ๐Ÿธ สœ แด แดœ ส€ ๊œฑ แด› แด ษข แด ! โ–‘โ–‘โ–‘โ–’โ–’โ–“โ–ˆ แดŠ แด ษช ษด แด› สœ แด‡ ส€ แด‡ แด› ส€ แด แด‡ ส€ แด€ โ–ˆโ–“โ–’โ–’โ–‘โ–‘
สŸ แด แด€ แด… ษช ษด ษข แด› สœ แด‡ สœ แด แด แด…... ๐ŸŸข ๐Ÿค แด„ แด สŸ สŸ แด€ ส™ s แด แด˜ แด‡ ษด ! ๐Ÿ” ส€ แด› + แด… ส€ แด แด˜ ส แด แดœ ส€ แดก แด€ สŸ สŸ แด‡ แด› ๐ŸŽฎ แด˜ สŸ แด€ ส แด› สœ แด‡ ส™ แด‡ แด› แด€ ษข แด€ แด แด‡ ๐Ÿ‘พ แดŠ แด ษช ษด แด› สœ แด‡ 8 ส™ ษช แด› s สœ แด แด แด… แดก แด€ ษด แด› แด› แด แด„ แด สŸ สŸ แด€ ส™ ? แด… แด แดœ s ๐Ÿ“ฉ
11
11
13
261
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
Absolutely! Real progress starts at the foundation, not just with layers built around it. When you improve the core, everything connected to it becomes stronger, smarter, and more efficient. That's the kind of fundamental innovation that creates lasting impact. Keep building! ๐Ÿค˜๐Ÿ’ช๐Ÿ”ฅ @BlockchainDan
Replying to @BlockchainDan
Everyone so focused on adding things around the model to improve. This is the low-level work that creates baseline performance improvements. Everything you slap on top is improved via the transitive property. ๐Ÿค˜๐Ÿ’ชโœจ
2
3
35
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
I own this masterpiece โค๏ธโ€๐Ÿ”ฅ
94
31
5
182
2,459
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
Everyone so focused on adding things around the model to improve. This is the low-level work that creates baseline performance improvements. Everything you slap on top is improved via the transitive property. ๐Ÿค˜๐Ÿ’ชโœจ
2
3
1
7
117
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
โ–‘โ–‘โ–‘โ–’โ–’โ–“โ–ˆ ษข ษช แด  แด‡ แด€ แดก แด€ ส แด› ษช แด แด‡ โ–ˆโ–“โ–’โ–’โ–‘โ–‘โ–‘ ๐ŸŽ ๊œฐ ส€ แด‡ แด‡ แด ษช ษด แด› ๊œฑ แด˜ แด แด› ๊œฑ ๐ŸŽ แดก แด‡ ' ส€ แด‡ ษข ษช แด  ษช ษด ษข แด€ แดก แด€ ส ๊œฐ ส€ แด‡ แด‡ แด ษช ษด แด› ๊œฑ แด˜ แด แด› ๊œฑ ๊œฐ ส€ แด แด ๐Ÿพ ส™ ษช แด› ๊œฑ สœ แด แด แด… ! ๐ŸŽฎ ๐Ÿพ ส™ ษช แด› ส€ แด‡ แด› ส€ แด แด‡ x แด˜ แด‡ ส€ ษช แด‡ ษด แด„ แด‡ แด ษด ส€ สœ โžœ ๊œฐ แด สŸ สŸ แด แดก @8bitshood and @gtfu_n_grind โžœ สŸ ษช แด‹ แด‡ โค๏ธ โžœ ส€ แด‡ แด˜ แด ๊œฑ แด› ๐Ÿ” โžœ แด… ส€ แด แด˜ ส แด แดœ ส€ แดก แด€ สŸ สŸ แด‡ แด› ๐Ÿ‘‡ โณ ๐Ÿท ๐Ÿธ สœ แด แดœ ส€ ๊œฑ แด› แด ษข แด ! โ–‘โ–‘โ–‘โ–’โ–’โ–“โ–ˆ แดŠ แด ษช ษด แด› สœ แด‡ ส€ แด‡ แด› ส€ แด แด‡ ส€ แด€ โ–ˆโ–“โ–’โ–’โ–‘โ–‘
๐Ÿน,๐Ÿน๐Ÿน๐Ÿน แดœษดษชวซแดœแด‡ แด‡xแด˜สŸแดส€แด‡ส€s ๐Ÿท๐Ÿผ แด›ส€แด€ษชแด› สŸแด€สแด‡ส€s ๐Ÿผ ส€แด€ส€ษชแด›ส แด›ษชแด‡ส€s แด…ส€แดแด˜ แด‡แด แด ๐Ÿ‘พ สแดแดœ แดษชษขสœแด› ส™แด‡ ๐Ÿพส™ษชแด›s ๐Ÿ‘พ
22
19
1
24
709
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
I understand ~0.001% of this paper; love every bit. Awesome when shares like this pop up. It's a difficult & critical side of AI most users entirely unaware of. People forget just how deep the rabbit hole goes ๐Ÿ‡ This is what REAL bare metal AI looks like ๐Ÿ”ฅ๐Ÿ’Ž. Keep pushing!
Great paper on why LLMs don't need Transformers when they write model weights by hand. This paper made me rethink how much we take the Transformer for granted. We treat it like the default for everything, forgetting it mostly won because it plays nicely with gradient descent and dense GPU math. Hyunjong Ok and Jaeho Lee decided to test what happens when you throw backprop out the window and ask an LLM to write a modelโ€™s weights entirely by hand. To figure this out, they put Claude Opus 5.5, GPT-6.1 Sol, and GPT-6 Astra into a sandbox with just ten labeled examples. They compared two setups: forcing the LLMs to fill in weights for a fixed Transformer versus giving them a blank slate to invent their own architecture from scratch. When given total freedom, every single model ditched dense neural layers and independently built a sparse associative-memory table instead. As someone who spends way too much time debugging weird model behavior, this makes total sense. Without gradients, writing weights is basically just coding. If a dense matrix gets a question wrong, good luck figuring out which floating-point number to fix. With a plain-text table, every mistake points straight to one row, so you can patch or delete a bad fact without breaking everything else. The results they pulled off on the ARC science benchmark are pretty wild. Their LLM-designed models hit up to 54.4% accuracy, outperforming the best LLM-written Transformer while using roughly 1,800x fewer parameters. Even better, their associative-memory model runs up to 17x faster on GPUs than a same-sized Transformer. Theyโ€™re upfront that pretrained LLMs still beat this on raw accuracy overall, so it's not replacing frontier models tomorrow. Still, itโ€™s a really cool reminder that the best architecture for gradient descent isn't the best architecture when an LLM is the one writing the weights. FULL PAPER PDF: arxiv.org/pdf/2611.22055
3
4
1
9
360
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
This is what I tell people. Find a job. Even better if itโ€™s in Web3, where your obsession can compound. โ€œEvery single minute that I had of free time, I would spend it in CT.โ€ โ€œItโ€™s a job to fund your dream.โ€ @FF_V12 with @Maximusss ๐Ÿ‘‡ Locked in more than ever.
8
2
1
16
516
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
Salvor started as a fix for my own AI workflow. It's becoming something much bigger. Today I'm introducing Salvor Labs, Salvor+, and 5 ways $SALVOR will unlock premium services, agent-native payments, and priority access. The next chapter starts here โ†“
41
15
3
60
6,652
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
I tried to mimic @Iamdotify , but I ended up mixing up the tagline. Iโ€™ll do better next time๐Ÿฅบ
To be profitable trading, all you might need is a day job.
3
1
11
251
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
๐Ÿ†Sible Ambassador Program: Cycle 1 USDC Rewards Distributed! Another milestone achieved for Sible Network! Weโ€™re pleased to confirm that USDC rewards have been successfully distributed to the top-performing Ambassadors of Cycle 1, recognizing their outstanding contributions to the growth of our global community. This marks an important step in the Sible Ambassador Program, turning community participation into tangible recognition and demonstrating our commitment to delivering on the programโ€™s reward structure. Congratulations to our top Ambassadors! Your dedication and impact deserve the spotlight. Explore the Cycle 1 rankings below and celebrate the individuals who made this first cycle a success. Cycle 1 sets the foundation. Weโ€™re building for whatโ€™s next. #SibleAI #SIB #SibleNetwork
367
765
22
2,005
19,542
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
Giveaway FEW spots for @moto_the_shadow To enter: โ—† Follow @moto_the_shadow and @JSTK_NFT โ—† Like & repost โ—† Drop your EVM wallet below โณ Ends in 24 hours
41
32
59
684
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
Minted this jewel thank you @Soathefirstt and @craigscoinpurse for this appreciate ๐Ÿ™
1
6
85
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
Iโ€™ve been creating content for a while, and I thought itโ€™s time I properly introduced what I do and what I bring to the table. Iโ€™m a Web3 content creator. I create videos, edit, tell stories, develop content ideas and help projects communicate what theyโ€™re building in a way people understand. I enjoy taking an idea and turning it into something people want to watch, read and engage with. If youโ€™re a founder or part of a team looking for someone to help bring your ideas to life through content, Iโ€™d love to work with you. My DMs are open. Letโ€™s create something together.
4
5
13
239
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
GM GM CT ๐Ÿ•Š๏ธ Have a profitable day
37
12
1
43
504
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
GM CT โœจ๏ธ Weekend maxing
46
5
42
326
๐‘ด๐’€๐‘ฒ๐‘ฌ ๐‘ด๐‘จ๐‘ฎ๐‘ต๐‘ผ๐‘บ retweeted
A $50-$100 Paid mints always get to public phase no matter how hyped they are. Lol these peeps knows exactly what they are doing
1
4
54