Ben Lopata

[blockgraze @blockgraze]

"bro have you tried clawdbot it's so insane"

"a little but not much, what do you use it for"

"it's crazy many you can do anything with it"

"what are you doing with it"

"you gotta try it"

"try it for what"

"don't get left behind man"

[Ben Lopata @bengineer10]

I built https://t.co/q2O59ekqpA for anyone who has had the displeasure of parking in the city of chicago

[Ben Lopata @bengineer10]

@moondrencht Wilson

[Ben Lopata @bengineer10]

@samoyed_online why do all these dudes look like a ps5

[Ben Lopata @bengineer10]

@aidenybai soo much better

[Andrej Karpathy @karpathy]

More apps should natively offer this.

“Export for prompt” button

[Ben Lopata @bengineer10]

@boozecruiser694 @incognitoStunna entertaining for whom exactly

[ROSE LANI @pradamercury]

How it feels to ask someone if they’re gonna eat their pickle

[Andrej Karpathy @karpathy]

I don't have too too much to add on top of this earlier post on V3 and I think it applies to R1 too (which is the more recent, thinking equivalent).

I will say that Deep Learning has a legendary ravenous appetite for compute, like no other algorithm that has ever been developed in AI. You may not always be utilizing it fully but I would never bet against compute as the upper bound for achievable intelligence in the long run. Not just for an individual final training run, but also for the entire innovation / experimentation engine that silently underlies all the algorithmic innovations.

Data has historically been seen as a separate category from compute, but even data is downstream of compute to a large extent - you can spend compute to create data. Tons of it. You've heard this called synthetic data generation, but less obviously, there is a very deep connection (equivalence even) between "synthetic data generation" and "reinforcement learning". In the trial-and-error learning process in RL, the "trial" is model generating (synthetic) data, which it then learns from based on the "error" (/reward). Conversely, when you generate synthetic data and then rank or filter it in any way, your filter is straight up equivalent to a 0-1 advantage function - congrats you're doing crappy RL.

Last thought. Not sure if this is obvious. There are two major types of learning, in both children and in deep learning. There is 1) imitation learning (watch and repeat, i.e. pretraining, supervised finetuning), and 2) trial-and-error learning (reinforcement learning). My favorite simple example is AlphaGo - 1) is learning by imitating expert players, 2) is reinforcement learning to win the game. Almost every single shocking result of deep learning, and the source of all *magic* is always 2. 2 is significantly significantly more powerful. 2 is what surprises you. 2 is when the paddle learns to hit the ball behind the blocks in Breakout. 2 is when AlphaGo beats even Lee Sedol. And 2 is the "aha moment" when the DeepSeek (or o1 etc.) discovers that it works well to re-evaluate your assumptions, backtrack, try something else, etc. It's the solving strategies you see this model use in its chain of thought. It's how it goes back and forth thinking to itself. These thoughts are *emergent* (!!!) and this is actually seriously incredible, impressive and new (as in publicly available and documented etc.). The model could never learn this with 1 (by imitation), because the cognition of the model and the cognition of the human labeler is different. The human would never know to correctly annotate these kinds of solving strategies and what they should even look like. They have to be discovered during reinforcement learning as empirically and statistically useful towards a final outcome.

(Last last thought/reference this time for real is that RL is powerful but RLHF is not. RLHF is not RL. I have a separate rant on that in an earlier tweet
https://t.co/RMIpFPVpuM)

[ₕₐₘₚₜₒₙ @hamptonism]

> be an electric engineering student
> team up w/ cracked classmates
> start quant trading
*we’re so cracked*
> founded a quant firm in his 30’s
> makes ¥100B trading with ai/ml
*we’re even more cracked with ai*
> buys thousands of Nvdia GPUs
> creates DeepSeek as a side project

[Kpaxs @Kpaxs]

Brain plasticity is your biggest asset. Use it carefully.

[Tim Urban @waitbutwhy]

Good day to remember just how big that green tree is

[Sotiris Tsern @SotirisTsern]

@Seamus_Malek https://t.co/b9B7A195VY

[Ben Lopata @bengineer10]

at this point, we are paying rent at the ship yards 🛳️

[Courtland Leer @courtlandleer]

fresh honcho site:

https://t.co/7TtycQep5T

[Ben Lopata @bengineer10]

@alexalbert__ are you a developer or yourself an AI agent? If so, consider applying for a grant from @xenograntai powered by @plasticlabs and push the frontier of open source AI!

[Ben Lopata @bengineer10]

@YouSimDotAI /draw vikings-in-combat

[Ben Lopata @bengineer10]

@YouSimDotAI /enhance chill-guy-status-of-bears --drawing=hyperstitional