i know it feels like you missed $HYPE but you probably felt that same way at $10 a few years ago, at $35 last year and at $45 earlier this year when i shilled it to my dad, he had never heard of it and i just told him it's great and i think it will go up a lot, he bought being aware early to something and not buying has cost ppl more money than being late it's why a lot of ppl who have known about crypto for a decade are still working day jobs. just knowing about something does not translate to buying and holding, it quite often translates to the opposite the ppl who are most angry about btc at the top are those who have been aware of it since $100 but never bought, they need to self-justify their actions by continuing to believe it's a scam. taunting themselves with fictional tales of what if you should not conclude a token's value based on your previous experiences with it, you should try to forget everything before today and understand if it is likely to go up you need to remember how to dream of brighter days not rue what could have been if you did things differently today is a day you can do things differently, so that in the future you don't have to think about it today will someday be what could have been

Cold start liquidity, market maker extraction, and inefficient OTC settlement have held back many of today's onchain derivatives. @variational_io has the answer to all three. We spoke with CEO @variational_lvs to hear all about it in this interview: defillama.com/research/inter…

You just won the strangest lottery: 8 weeks in Italy every year, for life, fully paid. One condition. You pick a single region now and can never set foot in another. Where are you spending forever?

Watch me control my computer with just my voice. This is the future of operating systems. No hands. GPT-Realtime 2.0 is very, very underrated. Demo:

Shuffle it

Announcing NVIDIA DLSS 5, an AI-powered breakthrough in visual fidelity for games, coming this fall. DLSS 5 infuses pixels with photorealistic lighting and materials, bridging the gap between rendering and reality. Learn More → nvidia.com/en-us/geforce/new…

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. github.com/karpathy/nanochat… 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.