Alex 🍋❄️.ETH

990 posts

Alex 🍋❄️.ETH banner
Alex 🍋❄️.ETH

Alex 🍋❄️.ETH

@itsalexey

In the grim darkness of the far future, there is only war and client diversity. Catch deepfakes with @BitMind, it is cool!

Decentral Katılım Kasım 2019
1.3K Takip Edilen2.1K Takipçiler
Sabitlenmiş Tweet
Alex 🍋❄️.ETH
Alex 🍋❄️.ETH@itsalexey·
Thrilled for this chance with @Axelarcore & @easya_app & @harvard_crypto! Congrats to fellow winners & participants. Hacking with @canvi_eth on Warp Drive was a blast! Play and test at wdrive.io. Let's take multi-chain to mainnet! @Stephenfluin inspired the idea!
EasyA 🤳📱@easya_app

🔥 Our amazing @Axelarcore winners ($20,000): 🥇 Warp Drive (Texas A&M) 🥈 CookieMart (Northeastern) 🥉 LemonSoda DAO (Northeastern) 💫 BlockXfer (Northeastern) 💫 Axinsure (National University of Singapore)

English
26
26
62
21.3K
Alex 🍋❄️.ETH
Alex 🍋❄️.ETH@itsalexey·
@namcios Can anthropic stop “killing” everything every day, pleez? Or is this “killing” just daily engagement farming?
English
0
0
1
175
Felipe Demartini
Felipe Demartini@namcios·
A Anthropic acabou de matar o Markdown. Um engenheiro do Claude Code publicou um artigo ontem que pode decretar o início de uma nova era. A tese é brutal: Markdown nunca foi o formato certo para comunicação entre humanos e IA. Era só o que tínhamos. O próprio autor admite que nunca leu um arquivo Markdown gerado por IA com mais de 100 linhas até o fim. Você também não lê. Eu também não. A sacada: Markdown assume que você vai ler do início ao fim. HTML assume que você quer ver o que importa e mexer com as mãos. Na prática: → 30 tickets de projeto viram kanban arrastável com colunas Now / Next / Later / Cut e botão de exportar → Lógica de rate limiting vira flowchart SVG com código inline, no lugar de 200 linhas de texto → Code review vira diff colorizado com grafos de dependência entre módulos → Parâmetros de animação, cores, regex, cron jobs ganham sliders com preview ao vivo → Specs de projeto viram 6 opções lado a lado com mockups interativos Todos exemplos reais do artigo. Todos substituem um muro de texto por algo que você de fato abre e usa. O trade-off existe: HTML é 2-4x mais lento para gerar. Mas com contexto de 1 milhão de tokens, esse custo sumiu. E a parte que ninguém está discutindo: o HTML gerado não é só para humanos. O agente de verificação também lê. O spec deixou de ser documento e virou memória compartilhada entre agentes. Markdown é relatório. HTML é interface. Relatórios são para ler. Interfaces são para continuar o trabalho. Se você usa IA em 2026 e ainda pede Markdown para tudo, você pode estar usando um smartphone como lanterna.
Thariq@trq212

x.com/i/article/2052…

Português
168
171
2.3K
1.2M
Alex 🍋❄️.ETH
Alex 🍋❄️.ETH@itsalexey·
It's open source, MIT licensed, pip installable: github.com/ztsalexey/epoc… Run it on your own models: epoch-bench run --provider openai --model gpt-4o 320 hand-crafted questions. 12 model results included. 148 tests passing. Star it if you think benchmarks should measure reasoning, not memorization.
English
0
0
1
46
Alex 🍋❄️.ETH
Alex 🍋❄️.ETH@itsalexey·
Three things make EPOCH different from other benchmarks: 1. It's self-generating. A dependency graph extracts 330 tech nodes from the questions and procedurally generates unlimited new ones. Can't saturate it. 2. Every benchmark is secretly a contamination detector — if you design it with paired counterfactuals. 3. Scaling doesn't close the gap (p=0.40). More compute doesn't buy you counterfactual reasoning.
English
1
0
1
42
Alex 🍋❄️.ETH
Alex 🍋❄️.ETH@itsalexey·
I built EPOCH — a benchmark that catches LLMs cheating. Every question has a twin: one about real history, one about an alternative timeline. The gap between them measures memorization vs. actual reasoning. Results from 12 models are... revealing. 🧵 Thread:
English
2
0
6
90
Alex 🍋❄️.ETH
Alex 🍋❄️.ETH@itsalexey·
New llm benchmark just dropped. Starting with testing
Alex 🍋❄️.ETH tweet media
English
0
0
2
52
Ken Jon
Ken Jon@kenjon·
We @bitmind have been building SOTA deepfake detection systems since before the '24 elections. If we can augment your teams workflow to identify these images/videos, we would happy to help. Here's an example of BitMind at work: • Any media data uploaded to a db, loading in DOM can be evaluated against our API • Results sent back including inference classification, C2PA, similarity for known images, VLM analysis Heres a video showing how it provides real time classification when scrolling on X
English
7
6
53
5.7K
أَحْمَد حَمْدَان
صواريخ إيرانية وانفجارات في تل أبيب ليلة قوية، الله يسدد الرمي يارب 🔥🔥🔥🔥🔥
العربية
473
1.8K
9.7K
1.7M
$trong
$trong@StrongHedge·
.@cz_binance causing the crypto liquidation event ever 10/10 Torched over the timeline @binance meanwhile “We’re looking for a PR KOL BD” So out of touch with reality it’s almost ridiculous, lol Maybe don’t tweet “What’s a 10/10 chart?”
$trong tweet media
Binance@binance

We’re looking for a PR KOL BD Build KOL relationships, support events, and gain real PR experience in the crypto space. Fresh grads with strong communication skills this is your sign 👋 🔥 Apply now! jobs.lever.co/binance/b22f7d…

English
61
77
640
64.8K
Cointelegraph
Cointelegraph@Cointelegraph·
⚡ REKT: $131M in longs were wiped out in the past hour.
Cointelegraph tweet media
English
109
65
575
43.7K
Shiitake
Shiitake@KintsuShiitake·
After fading $TAO at $30 in 2023, even though I came very close to aping, I just bought my first one at $236. These Bittensor subnets are too cool. SN44 caught my attention with @webuildscore. Probably gonna DCA for the first half of 2026.
Shiitake tweet media
English
42
50
377
23.2K