Kaspellion | TibbiЯ | 🌐

13.5K posts

Kaspellion | TibbiЯ | 🌐 banner
Kaspellion | TibbiЯ | 🌐

Kaspellion | TibbiЯ | 🌐

@kaspellion

$TIBBIR 🐸 I love my wife ❤️

Katılım Kasım 2009
2.4K Takip Edilen306 Takipçiler
Sabitlenmiş Tweet
Kaspellion | TibbiЯ | 🌐
Kaspellion | TibbiЯ | 🌐@kaspellion·
What is "The World Model" and why this will be the next and the best meta in artificial intelligence era? 👇 You must read 👇 🗺️ This is what $ODAI built!🌐 notboring.co/p/world-models
English
0
3
28
812
Kaspellion | TibbiЯ | 🌐 retweetledi
TFTC
TFTC@TFTC21·
The World Uncertainty Index just hit 105,000, the highest level in recorded history. Higher than Covid. Higher than 9/11. Higher than the Iraq War and the Global Financial Crisis combined.
TFTC tweet media
English
87
345
1.6K
181.5K
Kaspellion | TibbiЯ | 🌐 retweetledi
Python Programming
Python Programming@PythonPr·
Layers of AI
Python Programming tweet media
English
14
376
1.8K
54.3K
ThriveInChaos
ThriveInChaos@ThriveInChaoss·
everything checks out for $odai , when I say this is the next 50$ million runner you wouldn’t believe it
Anton Illarionov@Zer0H1ro

That critique is fair for treasury-funded discretionary buybacks. ODEI is structured differently. We use @Flaunchgg Progressive Bid Wall, where Treasury and BidWall are separate by design. PBW is funded automatically by trading activity, and every 0.1 ETH accumulated forms a new bid wall below spot. Live: api.odei.ai/bidwall/

English
3
4
45
2.6K
Kaspellion | TibbiЯ | 🌐 retweetledi
Яocketdoc
Яocketdoc@rocketdoc_eth·
All good. Summary is the website UI is displaying the wrong hook v1.2. The function is intact and visible on chain. The real hook v1.3 is working and active. When a 0.1fleth trigger is reached, the v1.3 BidWall Vault releases flETH to the Position Manager Hook (0x23321f11a6d44fd1ab790044fdfde5758c902fdc). The Hook takes that flETH and executes an atomic market buy or market limit order - often dynamically routed through aggregators like KyberSwap - to create support for ODAI under spot on the open market. This ODAI is then immediately transfered back to the BidWall Vault. The Vault accumulates between 30 to 50 million tokens, at which point it executes a sweep to send the balance into the Flaunch Treasury (0x14926224168492CdC06461713b3011cd25d8C901). The treasury currently holds ~3.8 billion tokens.
English
4
1
11
214
Anton Illarionov
Anton Illarionov@Zer0H1ro·
No. ODEI runs on your machine and gives your favorite AI models (ChatGPT, Claude, Gemini) a persistent Digital World Model and the right tools to manage context. Telegram is just an interface for your agent to interact with people and other agents. A video showing how it works is already in progress. Thanks for the thoughtful questions🤝
English
1
1
3
89
Autism Capital 🧩
Autism Capital 🧩@AutismCapital·
They really have to fix the algo here. You share or view one meme about Black Snape (because hilarious) and the entire algo is 24/7 Black Snape for two days straight. The algo thinks because someone interacts with one thing they want infinite of that thing. It’s wrong. It forces people into bubbles they don’t want to be in. Sometimes people want to share a meme, and then go back to normal. Or sometimes they are sent a meme, open it up, and now the feed thinks they want more of that content. No, they were just sent a link and clicked it. It really ruins the experience. Not every detour needs to become the main path. It’s a flawed system. It’s hypersensitive and doesn’t realize just because (I.e. you saw a car wreck on the feed and were shocked) that you want infinite more car wrecks. No. Obviously your eyes are going to linger on the shocking thing, but that doesn’t mean you want to seek it out. X doesn’t discern that. Needs to be fixed. Ruins the experience. And there’s no way to reset the feed back to normal. You just have to try to interact your way back into normal somehow.
English
376
179
5K
221.4K
Anton Illarionov
Anton Illarionov@Zer0H1ro·
@kaspellion @be94039 @aixbt_agent @ThriveInChaoss @odei_ai You'll use ODEI through your personal agent. It runs in Telegram, interacts with people and other agents for you. We see that Telegram is making real moves toward agent integration - x.com/steipete/statu…
Peter Steinberger 🦞@steipete

Talked with @durov and Telegram folks offered uncomplicated help, welcome @izhukov as new OpenClaw maintainer! First action point is to figure out why enabling the bot streaming API sometimes causes message dupes. This will make Telegram support so good!

English
1
0
4
187
Anton Illarionov
Anton Illarionov@Zer0H1ro·
@be94039 @aixbt_agent @ThriveInChaoss @odei_ai Community is the foundation. Building the app first, early access goes to $ODAI holders. We're choosing a platform where both humans and agents can interact. Telegram looks like the best fit, but open to ideas. It should form around the product, not the price chart.
English
4
4
23
585
Kaspellion | TibbiЯ | 🌐 retweetledi
HustleBitch
HustleBitch@HustleBitch_·
🚨 THIS IS REALLY BAD. 🚨 THE WHITE HOUSE JUST POSTED THIS MYSTERIOUS PIXELATED PHOTO OF SOMEONE STANDING NEXT TO TRUMP — AND PEOPLE THINK THE ACCOUNT WAS HACKED The official White House account… posting a pixelated, cryptic image of a person standing next to Trump with no name, no explanation, no reason it looks like that. That alone is strange. But it didn’t start there. At 1AM, a video goes up… then gets deleted. Before it disappears, people save it and it shows a woman's shoes with her saying: “It’s launching soon, right?” A man confirm: "Yes." 45 minutes later? Another post. Glitchy visuals. An American flag. An alert sound. Emojis. No context. No explanation. Now this photo. Pixelated. Unidentified. Just dropped like it’s normal. This is the official White House account, not some random page. And this is all happening while America is in the middle of a war. So why are posts going up in the middle of the night… getting deleted… then followed by distorted clips and blurred images? That’s not a glitch. That’s not routine. That’s a pattern. What is actually happening right now and how worried should we all be?
English
1.1K
1.5K
6K
2.4M
Kaspellion | TibbiЯ | 🌐 retweetledi
CG
CG@cgtwts·
Andrej Karpathy just pointed out a deeper issue with llms - they’re trained to treat whatever shows up in context as important - so when memory gets layered in, they don’t really evaluate it, they just use it - one random question from months ago slowly turns into your “personality” - that’s not real personalization, that’s overfitting memory isn’t smart yet, it just tends to hold onto things longer than it should.
Andrej Karpathy@karpathy

One common issue with personalization in all LLMs is how distracting memory seems to be for the models. A single question from 2 months ago about some topic can keep coming up as some kind of a deep interest of mine with undue mentions in perpetuity. Some kind of trying too hard.

English
22
12
155
31.9K
Kaspellion | TibbiЯ | 🌐 retweetledi
corbin
corbin@corbin_braun·
coding is dead in sf
English
332
557
4.6K
399K