CryptoPës

[Robot James 🤖🏖 @therobotjames]

idiot edge

sometimes edges appear due to other trader's stupidity

i'll show you a crypto trade that is so silly you might not believe it

but i promise you it's real.

[St. Hubert @ArbStHubert]

This remains true for global, top_positions, top_accounts ratios.

When combined (glb+pos+acc)/3 we can get a pretty decent feature family though transformations and scaling I found to be delicate.

[Andrej Karpathy @karpathy]

Three days ago I left autoresearch tuning nanochat for ~2 days on depth=12 model. It found ~20 changes that improved the validation loss. I tested these changes yesterday and all of them were additive and transferred to larger (depth=24) models. Stacking up all of these changes, today I measured that the leaderboard's "Time to GPT-2" drops from 2.02 hours to 1.80 hours (~11% improvement), this will be the new leaderboard entry. So yes, these are real improvements and they make an actual difference. I am mildly surprised that my very first naive attempt already worked this well on top of what I thought was already a fairly manually well-tuned project.

This is a first for me because I am very used to doing the iterative optimization of neural network training manually. You come up with ideas, you implement them, you check if they work (better validation loss), you come up with new ideas based on that, you read some papers for inspiration, etc etc. This is the bread and butter of what I do daily for 2 decades. Seeing the agent do this entire workflow end-to-end and all by itself as it worked through approx. 700 changes autonomously is wild. It really looked at the sequence of results of experiments and used that to plan the next ones. It's not novel, ground-breaking "research" (yet), but all the adjustments are "real", I didn't find them manually previously, and they stack up and actually improved nanochat. Among the bigger things e.g.:

- It noticed an oversight that my parameterless QKnorm didn't have a scaler multiplier attached, so my attention was too diffuse. The agent found multipliers to sharpen it, pointing to future work.
- It found that the Value Embeddings really like regularization and I wasn't applying any (oops).
- It found that my banded attention was too conservative (i forgot to tune it).
- It found that AdamW betas were all messed up.
- It tuned the weight decay schedule.
- It tuned the network initialization.

This is on top of all the tuning I've already done over a good amount of time. The exact commit is here, from this "round 1" of autoresearch. I am going to kick off "round 2", and in parallel I am looking at how multiple agents can collaborate to unlock parallelism.
https://t.co/WAz8aIztKT

All LLM frontier labs will do this. It's the final boss battle. It's a lot more complex at scale of course - you don't just have a single train. py file to tune. But doing it is "just engineering" and it's going to work. You spin up a swarm of agents, you have them collaborate to tune smaller models, you promote the most promising ideas to increasingly larger scales, and humans (optionally) contribute on the edges.

And more generally, *any* metric you care about that is reasonably efficient to evaluate (or that has more efficient proxy metrics such as training a smaller network) can be autoresearched by an agent swarm. It's worth thinking about whether your problem falls into this bucket too.

[Armv7l @Armv7lFx]

New article is out on how i drained 5 fig from Paradex (And abused their API), with more technical explanation, lot of footage.

for those who wonder what i do on daily basis, and are curious and want to learn more, or get dopamine

https://t.co/8rQ4Wppsek

[CryptoPës @CryptoSobaca]

MONAD BOXES 🎁

[CryptoPës @CryptoSobaca]

I just minted $VLTZ.
A pathway to $REYA.
Ethereum’s trading layer - s00n.
Check if you’re eligible → https://t.co/kyTMi4dss0
https://t.co/DxXNUEn02u

[Reya @reya_xyz]

Honouring the Voltz community

As we prepare for $REYA, we’re recognising one of the early pioneers of onchain finance: Voltz, and the community that helped move DeFi forward. 🧵

[jaehaerys 🪽 @0xJaehaerys]

analytics dashboard for @Lighter_xyz

lighter[.]gelora[.]study

you may find something useful and interesting

[aadvark @aadvark89]

all stats for @Lighter_xyz have 3-4xd since this (most importantly, OI)

and no, i'm not saying Lighter is the next Hyperliquid

still, its cool to see it as the #2 perp dex, while still in beta and with so much more shipping to do

points have a bid of $8 per point already but no sellers at this price

[Ralph Sueppel @macro_synergy]

Quantamental docs:
"Money growth is no longer carefully monitored, but it has been a pervasive and significant predictor of duration returns over the past 25 years. This holds true for both developed and emerging markets, as well as directional and relative returns, and across various sub-samples."
https://t.co/mKdPPJlJ8i

[CryptoPës @CryptoSobaca]

Yap early, yap only, yap often.

@_kaitoai is connecting AI, attention and capital with Yaps.

Just claimed my social card and I'm accumulating Yap points in real-time.

Claim yours 👉 https://t.co/uZOEVvXBSB

[Lookonchain @lookonchain]

July 5 Update:

9 ETFs added 166 $BTC($9.23M).
https://t.co/f0bGLFFovs

[Pixelverse @pixelverse_xyz]

Pixelverse, the emerging game ecosystem and entertainment studio, raised $5.5M from top VCs and Web3 founders.

Investors include @Delphi_Digital , @MeritCircle_IO , @MechanismCap , @BitscaleLabs and more, along with angel investors like @borgetsebastien , @dingalingts , @graildoteth and @LucaNetz

[CryptoPës @CryptoSobaca]

@eigencloud @eigenfoundation @eigenlabs Just a scam of community – breaking pendle contracts. Get no points on PT and YT in the first season – then who got those points from eth deposited on pendle. Looks like insider trading scam!!!

[ekri @ekrii3]

Feeling super generous so, somewhat reluctantly, here are 5 industry secret holy grail books that have immensely assisted me in my quantitative trading endeavours (as an ex manual HFT mm) that I highly recommend new and experienced traders pick up if they can. No specific order.

[EigenCloud @eigencloud]

Reminder - @eigencloud @eigen_da mainnet is coming soon! 🤠

Prepare yourself with this ELI5* pack:

EigenLayer Research Must Read:
- You Could've Invented EigenLayer: https://t.co/BzDDxeAOM8
- The EigenLayer Universe: Ideas for Building the Next 15 Unicorns: https://t.co/A32joNC93U
- Ecosystem thread https://t.co/KLYTLh1F4V
- Official youtube playlist https://t.co/a5enDmqh0R

On Eigen DA:
- Origin of EigenDA thread https://t.co/PB7rwb6mJb
- Intro to EigenDA Hyperscale Data Availability for Rollups. https://t.co/U9FtuzoWVc

(*A very advanced 5-yr old; h/t @0xtonkatsu)

[CryptoPës @CryptoSobaca]

I just claimed my AEVO airdrop!

Join with this referral link and earn 2 - 4x more AEVO on all trades 🎰

https://t.co/XTPneSGKuf

[ASXN @asxn_r]

1/

We've compiled a list of @EthereumDenver talks sorted by topic.

You can find it in the tweet below:

[Stat Arb @quant_arb]

Quick clarification, meant to say checked w PCA (scree plot) then you
fix it if it’s an issue (often it isn’t)

In general you shouldn’t do anything to your features after they’re done

That means no PCA. Keep it high dim. or fix your features

Same reasoning on LASSO dropping features, drop it from the start

[Jake @Jakeminlim]

YT Pricing Calculator for EigenLayer Airdrop

I’ve been doing some napkin math in my head for a 100% YT strategy on @pendle_fi.

If $EIGEN is $0.10 when airdropped, current YT eETH is selling at 34.5% discount and the profit from the strategy will net $1,250 per 1 ETH.

[Barchart @Barchart]

U.S. Banks are facing unrealized losses of roughly $685 billion (updated as of Q3). This problem isn’t going away any time soon until the Federal Reserve begins cutting. New York Community Bancorp $NYCB might be the next victim.