ZenChamp

242 posts

ZenChamp

ZenChamp

@Formant561

Присоединился Mayıs 2020
182 Подписки27 Подписчики
ZenChamp ретвитнул
Vadim
Vadim@VadimStrizheus·
pov: you have been using Claude Code for 3 months and just discovered skill graphs. Give this to your agent. You’ll thank me later. 👇
Heinrich@arscontexta

x.com/i/article/2023…

English
32
128
1.7K
385.7K
ZenChamp ретвитнул
Andrej Karpathy
Andrej Karpathy@karpathy·
Wow, this tweet went very viral! I wanted share a possibly slightly improved version of the tweet in an "idea file". The idea of the idea file is that in this era of LLM agents, there is less of a point/need of sharing the specific code/app, you just share the idea, then the other person's agent customizes & builds it for your specific needs. So here's the idea in a gist format: gist.github.com/karpathy/442a6… You can give this to your agent and it can build you your own LLM wiki and guide you on how to use it etc. It's intentionally kept a little bit abstract/vague because there are so many directions to take this in. And ofc, people can adjust the idea or contribute their own in the Discussion which is cool.
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.

English
964
2.6K
25.2K
6.2M
ZenChamp ретвитнул
Michael Magán
Michael Magán@mrmagan_·
your app's UI has entered the chat. just register your UI components and your APIs. build an agent that speaks your interface in minutes 👇
English
52
44
858
105.5K
ZenChamp ретвитнул
LUKSO
LUKSO@lukso_io·
This is your go-to guide for AI agents using @ERC725Account on LUKSO. Here’s everything you need to know to give an agent an identity 🆙 with scoped permissions (LSP6), gasless transactions, and instant revocation. Agents are already live, connect yours 👇 medium.com/lukso/how-to-c…
English
6
30
114
13.8K
ZenChamp ретвитнул
eye zen hour 🥶
eye zen hour 🥶@eyezenhour·
$SUI Positioning Is Getting Interesting Here 👀 Across venues, liquidation pressure is skewed higher, not lower Full map context: ➭ ~$143.9M in shorts get liquidated if SUI pushes toward $1.74 ➭ ~$25.7M in longs get liquidated if price drops to ~$0.72 That’s nearly 6:1 upside pressure vs downside Now, isolate Binance: ➤ $16.18M short liquidations up to ~$1.25 ➤ $11.87M long liquidations down to ~$0.73 The skew narrows, but upside pressure still dominates near-term Hyperliquid adds another layer: → Shorts concentrated $1.15–$157 → Larger clusters above at $1.61, $2.22, $4.30, $6.70 → Longs relatively thin until ~$0.47 Translation: ➭ Shorts are layered into resistance ➭ Longs are not aggressively stacked below ➭ Downside forced selling exhausts faster than upside fuel Now review structure: ➭ Weekly at bottom of multi-month channel ➭ Pressing against Weekly Hypertrend ➭ StochastiX in oversold territory ➭ Sitting on a high-volume node This is compression at structural support while leverage stacks above TLDR: $SUI is in a zone where upside movement triggers more forced action than downside movement That does not guarantee direction But when liquidation asymmetry aligns with higher-timeframe support, volatility expansion usually follows If Hypertrend flips, positioning does the rest 🌊
eye zen hour 🥶 tweet mediaeye zen hour 🥶 tweet mediaeye zen hour 🥶 tweet mediaeye zen hour 🥶 tweet media
English
22
26
148
16.1K
ZenChamp ретвитнул
goodalexander
goodalexander@goodalexander·
reasons why crypto is down: 1. On chain stocks and bonds have not resulted in material fee capture for any major L1 protocol. Arbitrum pumped on Robinhood announcement only for Robinhood to build its own inhouse solution. Temporary Nasdaq on-chain trading pump then it turns out they're using private blockchains. SOL doing $1m a day of fees down from $24m Trump coin frenzy and $15m+ of inflation. 2. high inflation and primary equity story is international equities, gold, and AI. crypto is not getting enough attention 3. Trump promised mass deportation, less international engagement, low inflation. seems we are 0/3. With no Trump USD stablecoins and regulatory progress could reverse. Crypto is seen as Trump proxy. when Trump won EURUSD collapsed and BTC went bid. now EURUSD is surging while BTC is down 4. Gold / silver have captured 'sovereign diversification'. Russia appears to have successfully use gold to thwart US sanctions after invading Ukraine. Despite the fact it could have used crypto (it did not). Now China is looking at Gold as a good reserve asset if it chooses to invade Taiwan 5. "Digital Gold" story does not work if Gold is up only and AI is up only while coins are down only. Thus - people no longer believe. At the same time there are high valuations and relatively high network inflation 6. There has yet to have been material evidence of AI <> crypto inter-relation. Bittensor if anything by becoming more credibly non political is allowing subnets to do anything (not just AI related). Coin is dead. Narrative of AI agents doing things with crypto so far has not come to pass. 7. DATs, and MSTR buying at high prices are a double edged sword. As DATs drop below MNAV activists are incentivized to take stakes in DATs and sell the underlying tokens to drive shareholder value. In current market conditions this would be catastrophic 8. Maybe quantum fears - but I think we are memeing ourselves about that far more than ppl are actually trading on this I think it's important to understand these things. The market put on a huge bet that the system and crypto would mesh. That bet has imploded. It is not pretty. I got some things right and many things wrong. I am not planning on leaving the industry or pivoting to gold mining. Reasons to be excited about Crypto on a go-forward 1. People are not thinking through what a fully multi polar world looks like given current debt levels combined with CBDCs. base case is mass financial repression and wealth taxes targeted at individuals. Individuals cannot easily 2. There are only 100-200k researchers and engineers in the global AI industry. Unlike previous industrial revolutions - the growth in AI will not cause mass employment. It will instead cause mass unemployment. This will only accelerate confiscatory impulses and also put pressure on global central banks to ease (bullish fixed supply assets). AI itself will be effectively leveraged to track physical locations ala China making it difficult for individuals to effectively custody gold 3. Nation states may not get along but all of them will want to tax citizens 4. The original BTC run was Chinese high net worth individuals. We have a mental model of what this looks like. More repression. More capital controls. That's how BTC went from $100 -> $80k to begin with. 5. Unlike the original BTC run privacy technology has advanced considerably. ZK transactions are 20x cheaper than they were in 2021. Halo2 / Z cash have made substantive technical progress. Railgun is responsive to TornadoCash rulings and there are increasingly legal frameworks for privacy. 6. AI is fundamentally compression technology. Models are getting smaller. Open Source (like Kimi K2.5) is pretty solid and will remain subsidized by China indefinitely which is treating open Source AI development a bit like free marketing for CCP ideals. There will reach a threshold when there are models that are smarter than ppl that run on local hardware. This makes society portable in a way that was previously unimaginable. 7. The technical selling has 2 sources: a. DAT overhangs, which are reflexive and can expand to buying pressure if the market turns around b. Bloated teams, which should structurally be able to be cut due to AI coding 8. Re: quantum I think AI coding will also accelerate the speed at which some protocols become quantum resistant, and I imagine this will be faster than traditional websites. There are some undeniably scary 'quantum internet' bear cases out there but I think we're way too early for that Put another way: sovereign individuals were never really believable before "vibe coding". Because you'd need an army of employees. Who need national services. The economic viability of small scale technological economies is skyrocketing. And therefore even if traditional equities or bonds don't end up on public blockchains imminently there is still believably going to be substantial post-national on-chain volume. But it is coming at a lag vs market expectations, resulting in massive financial destruction Binance, Hyperliquid and Tron remain successful decade long experiments in - what are essentially equity like assets that exist completely outside of traditional jurisdiction or reporting requirements. It is fundamentally viable to run on-chain businesses, and will become more appealing over time bc of how messed up our world is I don't want to pretend that the current market condition is good or that "I Called it". or something. I am trying to make sense of it. If I had to sum it up "anti or post nation state assets bet the ranch on nation states and got screwed". Anarcho capitalists shocked their government partnership didn't work out. Makes sense in hindsight doesn't it.
English
94
93
1.1K
143.2K
ZenChamp ретвитнул
𝔸𝕟𝕥𝕙𝕠𝕟𝕪👾
Go to your ChatGPT and send this prompt: “Create an image of how I treat you”. Share your image result. 😂
𝔸𝕟𝕥𝕙𝕠𝕟𝕪👾 tweet media
English
18.7K
1.4K
21.4K
5M
ZenChamp ретвитнул
Boris Cherny
Boris Cherny@bcherny·
I'm Boris and I created Claude Code. I wanted to quickly share a few tips for using Claude Code, sourced directly from the Claude Code team. The way the team uses Claude is different than how I use it. Remember: there is no one right way to use Claude Code -- everyones' setup is different. You should experiment to see what works for you!
English
927
5.8K
50.9K
9.1M
ZenChamp ретвитнул
Tech Friend AJ
Tech Friend AJ@techfrenAJ·
Deployed @openclaw in under 5 minutes on AWS free tier. Open source personal AI. Full system access. Interfaces through WhatsApp, Discord, Telegram. People are rigging it to their Ray-Bans for real-time price comparisons. One command. That's it.
English
189
561
7.1K
1.5M
ZenChamp ретвитнул
Jelle
Jelle@CryptoJelleNL·
ok small correction
Jelle tweet media
English
3
6
74
4.2K
ZenChamp ретвитнул
Boris Cherny
Boris Cherny@bcherny·
I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently. So, here goes.
English
1.3K
7K
54.4K
8.1M
ZenChamp ретвитнул
ᴛʀᴀᴄᴇʀ
ᴛʀᴀᴄᴇʀ@DeFiTracer·
🚨 $BTC JUST HIT NEW ATH! As promised, @grok will pick 5 random comments to win $2,500 To join: like, repost and comment "me" 👇 Must be following to qualify, I'll check everything
ᴛʀᴀᴄᴇʀ tweet media
English
543
454
885
61.8K
Lofi The Yeti
Lofi The Yeti@lofitheyeti·
Something is brewing and this Yeti is ready to spill the alpha 👀🚀 Join us + @tradeportxyz tomorrow at 9am EDT / 1pm UTC as we cover what’s in store for Mystic Yeti holders and the Flurry 🥶🧊 twitter.com/i/spaces/1rmxP…
English
42
33
166
12.1K
Ika「🦑」
Ika「🦑」@ikadotxyz·
From Squid Ink to Yeti Stink.
English
78
40
306
15.4K
Lofi The Yeti
Lofi The Yeti@lofitheyeti·
Drop a 🥶 if you want a Yeti Funko...
Lofi The Yeti tweet media
English
186
48
328
9K
Lofi The Yeti
Lofi The Yeti@lofitheyeti·
SuiFest is 1 week away 🥶 Couldn't help myself, time to reward the Mystic Yetis ⤵️ 💎 Giving away $2,000 until SuiFest on 10/2 💎 Choosing 5 winners per day, $50 each 💎 Choosing one winner of $250 IRL at SuiFest How To Enter: 1️⃣ Own a Mystic Yeti NFT 2️⃣ Rock it as your 𝕏 PFP 3️⃣ Comment on our posts daily so we see 👀 Starts today 🚀 See you in Singapore in a week, legends 🇸🇬
English
172
76
290
21.8K