Omar pfp

Omar

@omarperacha.eth

89 Following
65 Followers


Omar pfp
Omar
@omarperacha.eth
A couple of pieces from my capsule of tokenised 1-of-1s 🍂
0 reply
0 recast
1 reaction

Omar pfp
Omar
@omarperacha.eth
Sticky toffee pudding is the GOAT dessert
0 reply
0 recast
0 reaction

Omar pfp
Omar
@omarperacha.eth
7 years ago I started this Generative AI project. I built and trained a Generative Adversarial Network (remember GANs?) to compose new music for the Japanese "shakuhachi" flute, after winning a competition by shakuhachi master Shawn Head. These were the first ever pieces of contemporary chamber music to be fully written by deep neural networks. Performance by Shawn Head recorded at Girton College, Cambridge University. https://www.youtube.com/watch?v=BubH5CVH6ug&list=RDBubH5CVH6ug&start_radio=1
0 reply
1 recast
5 reactions

Omar pfp
Omar
@omarperacha.eth
detail shots ✨ no.5028 from my capsule collection of tokenised 1-of-1s!
2 replies
1 recast
10 reactions

Omar pfp
Omar
@omarperacha.eth
Just read the "TradingAgents" paper by Tauric Research that's been trending Give different LLMs specific roles each (e.g. analyse reddit data, technical fundamentals, make the actual trade decision), and just coordinate them in an agentic framework. The crazy part for me? NO actual training/fine-tuning of the LLMs involved at all. They used various OpenAI models out of the box. I dug into the code and this one line in the prompt seems to encapsulate all iterative improvement: "Take into account your past mistakes on similar situations. Use these insights to refine your decision-making and ensure you are learning and improving... Here are your past reflections on mistakes" It's a purely agentic innovation. https://tauric.ai/research/tradingagents/
0 reply
0 recast
0 reaction

Omar pfp
Omar
@omarperacha.eth
Which palette/aesthetic from our Tokyo shoot is more memorable? A - burgundy/mahogany 🪵 B- purple/sakura 🌸
1 reply
0 recast
1 reaction

Omar pfp
Omar
@omarperacha.eth
Self-improving AI? 🤖 That's the claim by Sakana AI in their recent paper, "The Darwin Gödel Machine" - "AI that improves itself by rewriting its own code". I just finished reading the paper and it's not as scary as it sounds! They built an AI *Agent* that can edit its own code, but the underlying LLM (mostly Claude 3.5 Sonnet or o3-mini) stays fixed. Still a long way from AI actually inventing new ML architectures and training its own upgrade. We're safe - for now 😅
1 reply
0 recast
2 reactions

Omar pfp
Omar
@omarperacha.eth
Some BTS of our shoot in Tokyo a couple of months ago 🗼 Model wearing piece #5043 from my capsule of one-of-ones Sakura on Sakura 🌸
0 reply
0 recast
0 reaction

Omar pfp
Omar
@omarperacha.eth
Wearing the first FlowZed silk bomber. 1 of 1 🍂
1 reply
1 recast
2 reactions

Omar pfp
Omar
@omarperacha.eth
Importance of aligning a high quality silk satin with the direction of the weave 🧶 These two pieces of fabric are the exact same colour! One is correctly aligned with the warp yarn, the other with the weft instead 🌈 A fabric with a brilliant sheen like this silk exaggerates the mistake, scattering light so differently that the hues don’t visibly match
1 reply
0 recast
1 reaction

Omar pfp
Omar
@omarperacha.eth
Put some of my silk under an electron microscope 🔬 At 2000x magnification the fibres remind me of banyan trees I saw on a recent trip to Hong Kong
0 reply
0 recast
1 reaction

Omar pfp
Omar
@omarperacha.eth
👀 Loading in 2025… a capsule collection of token-backed, 1-of-1 silk outerwear pieces 🍂 Introducing FlowZed 1.0: The Merkle Tree
1 reply
0 recast
2 reactions

Omar pfp
Omar
@omarperacha.eth
Definitely not in Dubai anymore ⛈️
0 reply
0 recast
1 reaction

Omar pfp
Omar
@omarperacha.eth
"Mathematicians have long studied how shapes can fit together to cover surfaces without gaps. However, their typical approach – using shapes with sharp corners and flat faces – is rarely seen in the natural world. Instead, living organisms use a dazzling array of patterns to form and grow, for instance in muscle tissues. Most strikingly these patterns are characterised by shapes with curved edges, non-flat faces, and few, if any, sharp corners. Up to now, how nature achieves geometrical complexity using these ’soft shapes’ has eluded mathematical explanation. The answer... is a new class of mathematical shapes called soft cells. These shapes have a minimal number of sharp corners, and cover space without gaps." https://www.ox.ac.uk/news/2024-09-12-mathematicians-discover-new-universal-class-shapes-explain-complex-biological-forms#:~:text=A%20team%20of%20mathematicians%20from,sea%20shells%20to%20muscle%20cells.
0 reply
0 recast
2 reactions

Omar pfp
Omar
@omarperacha.eth
Biggest blockchain conference in the world but I still have to look after a badge AND wristband for a week 😂 “Don't lose your badge or wristband. Treat these with the same level of care as you would a passport.” 🤦🏽‍♂️ Proof of registration in a mobile wallet wen?
0 reply
0 recast
0 reaction

Omar pfp
Omar
@omarperacha.eth
Just got round to reading “The AI Scientist” paper by Sakana AI, along with teams in Oxford and Canada What I found most exciting about this work is that, unlike most recent AI breakthroughs in the current “scaling war” era, this one really could come from a team with modest resources. https://arxiv.org/abs/2408.06292
2 replies
1 recast
1 reaction

Omar pfp
Omar
@omarperacha.eth
https://onchain100-frame.vercel.app/api last week I introduced hybrid-chain NFTs as an effective way to decentralise dynamic NFTs on Ethereum, using /icp today I'm open-sourcing IC canister code to get started with your own hybrid-chain NFTs! https://github.com/omarperacha/eth-metadata-starter Example on @base.base.eth:
1 reply
0 recast
3 reactions

Omar pfp
Omar
@omarperacha.eth
Dynamic NFTs have a dilemma: ・Either you centralise metadata to make traits easy to change ・Or you store traits in the contract, but lose indexing by not returning them from the tokenURI Until now, thanks to /icp Free customisable (!) mint on @base.base.eth to explore for yourself https://onchain100-frame.vercel.app/api
1 reply
1 recast
3 reactions

Omar pfp
Omar
@omarperacha.eth
What’s the easiest way to do the following on ICP: Given a canister written in rust, call a method declared in its candid API via HTTP, passing arguments as query params? Do you need a frontend canister or can you directly call this rust “backend” canister?
0 reply
0 recast
0 reaction

Omar pfp
Omar
@omarperacha.eth
Question about ICP Canister ECDSA keys: If a canister can have an ECDSA key, does it follow that it effectively “has the Ethereum account” corresponding to that public key? I guess so since it can sign messages with the correct private key through threshold ECDSA - is there a step I’m missing?
2 replies
0 recast
0 reaction