Fitsu

[Andrej Karpathy @karpathy]

Judging by my tl there is a growing gap in understanding of AI capability.

The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.

But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.

So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.

TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.

[Andrej Karpathy @karpathy]

Thank you Sarah, my pleasure to come on the pod! And happy to do some more Q&A in the replies.

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[Andrej Karpathy @karpathy]

Farzapedia, personal wikipedia of Farza, good example following my Wiki LLM tweet.

I really like this approach to personalization in a number of ways, compared to "status quo" of an AI that allegedly gets better the more you use it or something:

1. Explicit. The memory artifact is explicit and navigable (the wiki), you can see exactly what the AI does and does not know and you can inspect and manage this artifact, even if you don't do the direct text writing (the LLM does). The knowledge of you is not implicit and unknown, it's explicit and viewable.
2. Yours. Your data is yours, on your local computer, it's not in some particular AI provider's system without the ability to extract it. You're in control of your information.
3. File over app. The memory here is a simple collection of files in universal formats (images, markdown). This means the data is interoperable: you can use a very large collection of tools/CLIs or whatever you want over this information because it's just files. The agents can apply the entire Unix toolkit over them. They can natively read and understand them. Any kind of data can be imported into files as input, and any kind of interface can be used to view them as the output. E.g. you can use Obsidian to view them or vibe code something of your own. Search "File over app" for an article on this philosophy.
4. BYOAI. You can use whatever AI you want to "plug into" this information - Claude, Codex, OpenCode, whatever. You can even think about taking an open source AI and finetuning it on your wiki - in principle, this AI could "know" you in its weights, not just attend over your data.

So this approach to personalization puts *you* in full control. The data is yours. In Universal formats. Explicit and inspectable. Use whatever AI you want over it, keep the AI companies on their toes! :)

Certainly this is not the simplest way to get an AI to know you - it does require you to manage file directories and so on, but agents also make it quite simple and they can help you a lot. I imagine a number of products might come out to make this all easier, but imo "agent proficiency" is a CORE SKILL of the 21st century. These are extremely powerful tools - they speak English and they do all the computer stuff for you. Try this opportunity to play with one.

[Etherex @etherexfi]

We flipped another epoch and hit $200M in TVL 🦖

Here’s a recap of what you might have missed this week:

[Proof of Narnian @real_narnia]

⚡Podcast September 5: Guest Vladimir Konovalov @1n5734d0fu from @HASHCIB
Topics: Science. AI. VC strategies and evaluation criteria. Trending narratives. $WLFI. $PUMP. $HYPE.
🎧 Link to the podcast: https://t.co/9GCAKjfb6o

[Ben Zhou @benbybit]

Join us on war against Lazarus - https://t.co/EEkj05V2xZ
Industry first bounty site that shows aggregated full transparency on the sanctioned Lazarus money laundering activities. V1 includes:
- Becoming a bounty hunter by connecting your wallet and help tracing the fund, when your submitted bounty leads to freeze, bounty is paid upfront upon instantly at freezing.
-All freezer gets 5% of the bounty, exchange, mixers and all.
- live ranking of good and bad actor and their response time to deal with the sanctioned Lazarus group transactions. You don't want to end up on the bad actor list , it's a record of you helping to facilitate sanctioned transactions.
- Live API wallet address update for exchange, Chainanalysis, @arkham @elliptic @trmlabs

We have assigned a team to dedicate to maintain and update this website, we will not stop until Lazarus or bad actors in the industry is eliminated. In the future we will open it up to other victims of Lazarus as well.

V2 coming up:
-Live update on latest wallet with latest balance, this will serve as open bounty, bounty hunter can claim a wallet and own that part of the journey for x amount of time when the wallet is moving.
-Regulator Tools
-
**More idea, feedback welcomed, please leave at comment, a lot more improvement needed**

LET THE HUNTING SEASON BEGIN!

[Foxy 🦊 @FoxyLinea]

Foxy’s face when he realizes the stars are aligning.

$FOXY monthly airdrop for NFT holders based on rarity 🦊

@MetaMask debit card cash back rewards 💰

@LineaBuild $LINEA airdrop

Should NFT communities receive a hefty airdrop? 🧐

FOXY FEBRUARY

[Monad @monad]

Monad Testnet is live.

Try it out here: https://t.co/QoJwPxRBTC

[SolanaFloor @SolanaFloor]

🚨BREAKING: 🇺🇸 The SEC has officially acknowledged the following Spot @Solana ETF filings today:

•Canary Solana ETF
•@vaneck_us Solana ETF
•@bitwise Solana ETF
•@21Shares Solana ETF

[Abstract @AbstractChain]

Abstract Mainnet.

Coming January 2025.

[Sergey (ICO Drops) @sergeydrops]

Other Low IQ bets
• ETH TREND: $EIGEN, $ZK
• AI TREND: $RENDER, $IO, $GRT
• BTC TREND: $STX, $CORE
• P2E TREND: $IMX, $BEAM
• SOLANA TREND: $SOL
• RWA: $ONDO, $CPOOL
SLOWLY BUT SURELY: $MNT, $BNB, $BTC, $DOGE

I don't see much point in overthinking and trying to outsmart the market. The ICOs are coming soon, and that's where we'll play smart. For now, let's just ride the trends.

[Sergey (ICO Drops) @sergeydrops]

Do you feel the latest narrative?

VCs are evil, just here to dump

The community is good - the community believes in a bright future and an eternal bull run

It's the perfect timing for startups to go for Public ICOs instead of Series A, raising $100-300 million and go-to-market without unlock.

Soon, the US will lift restrictions on ICOs. For the first time, Americans will be able to participate in ICOs properly. They want their 2017 back!

P.S. @ICODrops traffic is up 3x

[CZ 🔶 BNB @cz_binance]

gm, the food taste so good... And what a luxury to be able to have more than one piece of fruit per day!

I know some of you may have a lot of questions. I won't have all the answers.

Let me chill for a bit. Then figure out the next steps. There are always more opportunities in the future than there were in the past.

I want to thank everyone for your support. It meant a lot to me, and kept me strong in the darkest moments.

A few quick updates/thoughts:

Giggle Academy has been going well, and will be a big part of my life for the next few years.

Will continue to invest in blockchain/decentralized technologies, AI, and biotech. I am a long term investor who care about impact, not returns.

I will also dedicate more time and funding to charity (and education). I have some rough ideas.

Still working on my book. About 2/3 done, I think. Writing a book is a lot more work than I anticipated, but will see this one through.

Oh, @binance seems to be doing well without me back-seat-driving, which is excellent. Every founder's dream!

Stay tuned. See you at the conferences.

[CZ 🔶 BNB @cz_binance]

gm

[Fitsu @f1tsu]

I just claimed $BLAST from @BLUR_io's Season 3 airdrop. Season 4 starts today, and it's powered by @BLAST_L2, The Fullstack Chain.

[Fitsu @f1tsu]

I earned 61,000 $BLAST in @BLAST_L2's Phase 1 airdrop. Join the mission to UNBANK THE BANKED and earn Native Yield + BLAST in Phase 2!

[Magic Eden 🪄 @MagicEden]

As promised, Magic Tickets are getting integrated into our rewards program.

We’ve been removing Diamonds from inorganic users who have been clearly abusing the program, and today we’re reallocating them to you. You can now burn your Magic Tickets for Diamonds on Magic Eden.

The burn window lasts one month — Burn here: https://t.co/fb4WtZhzj2

[Dan Reecer @danreecer_]

✍️ Long-form tweet on the significance of today's Staking for Governance launch for @wormhole, what it means, how you can get involved, and what's to come:

What went live today:
As step one in the W staking roadmap, Staking for Governance is now live as the industry's first multichain governance system (MultiGov) begins to roll out.

W holders can now delegate their tokens to themselves (similar to 'registering to vote') or they can delegate their tokens to any Delegate (myself included) if they want to give their voting power to someone else given the time and/or expertise required to vote on governance proposals.

What is MultiGov?
Until Wormhole, @tallyxyz, and @ScopeLift built MultiGov, no protocol has been able to provide a seamless multichain experience for voting or delegating in any DAO. @Uniswap, for example, hosts its governance on @Ethereum mainnet so $UNI holders can only vote on proposals on Ethereum (expensive, and inconvenient for users who prefer other chains like @arbitrum or @base). Uniswap uses Wormhole to pass those governance decisions cross-chain to all their other deployments, but the governance decisions themselves are not multichain.

MultiGov, launching first with the @wormhole DAO, will leverage multichain token transfers, hub-and-spoke MultiGov tech built by the three teams previously mentioned, and Wormhole Messaging to allow users to delegate and vote from any connected L2 and soon Solana (other chains/VMs can be added in the future too). This is incredibly motivating to see real innovation in the space that is making true multichain experiences and chain abstraction a reality.

For a more technical deep-dive on how all of this works, check out @RobinsonBurkey's thread: https://t.co/JnVV95vjXM

Integrate MultiGov for your DAO: DAOs of several other top protocols are in the queue waiting to implement MultiGov for their community. Please DM me or message your contacts who contribute to Wormhole, Tally, or Scopelift so we can start working on the MultiGov implementation for your DAO. You can also submit this form if you'd like to contact the team to integrate MultiGov: https://t.co/E7tyFQ7hBv

How to Stake for Governance
Head to the Tally Governance Portal (https://t.co/i0q15iIBpT) and follow the process to transfer your W tokens from Solana to a supported EVM chain including Ethereum, Arbitrum, Optimism, and Base (if they are not already), choose your delegate (you can delegate to me if you'd like), and then stake your W for governance.

Next steps for Wormhole Governance
As mentioned, this is just the first step in Wormhole DAO Governance going live with this DAO's implementation of MultiGov. Over the coming weeks and months, the Solana integration into MultiGov will be completed/audited, and eventually Wormhole governance will begin accepting proposals and begin voting. More to come as well on updates to the staking roadmap.

Multichain is the future, and @wormhole is building it. Let's keep pushing.

[Wormhole @wormhole]

Stake For Governance is now live for W token holders!

This is your opportunity to participate in the governance and impact the future direction of the Wormhole DAO and platform.

Get started now at the @tallyxyz Governance Portal: https://t.co/U8QJdt84YX