We're launching a $100,000 challenge to see who can make Muse Spark run fastest: openkernelchallenge.dev/
It was an honor to bring two chapters of my life together with Meta's Muse Spark coming to Apple's Mac Studio. I hope folks enjoy it.
I am building out Meta's compute desk. I am looking for folks with background in any of: deep learning, supply chains, commodities, semiconductors, sovereigns, energy, Excel, prediction markets, monitoring situations, etc. Candidates should have experience in at least one of these domains, and novel ideas on how to apply LLMs to their work. If you'd like to help manage $100b+ of compute spend, please DM and include information about why you're a fit.
(If you're in doubt about being a good fit please apply! Epistemic humility is in short supply these days.)
Ilya and Daniel have assembled an incredible team and I'm honored to have been able to assist in getting SSI off the ground. The company's future is very bright, and I expect miracles to follow.
I sent the following message to our team and investors:
—
As you know, Daniel Gross’s time with us has been winding down, and as of June 29 he is officially no longer a part of SSI. We are grateful for his early contributions to the company and wish him well in his next endeavor.
I am now formally CEO of SSI, and Daniel Levy is President. The technical team continues to report to me.
You might have heard rumors of companies looking to acquire us. We are flattered by their attention but are focused on seeing our work through.
We have the compute, we have the team, and we know what to do. Together we will keep building safe superintelligence.
Ilya
.@ilyasut's consistent track record of revolutionary breakthroughs:
NVIDIA's Rubin represents a question: which do you expect to grow faster -- the demand for intelligence (inference) or the efficiency of the refinery (GPU)?
I think I want a little sidebar on my Mac that's transcribing everything it's seeing and hearing into possible queries. I click the query, it sends to Perplexity/OpenAI/Claude. Would that be useful? (It does not store transcripts or screenshots.)
Just passed $1b of contract value on gpulist.ai
Just passed $600m of contract value on gpulist.ai
Demoscene clips make for very humbling statements about the true Kolmogorov complexity of reality:
vec3 f=vec3(.2,.4,1.5),p;for(float i,e,g,S,u;i++<50.;o+=exp(e*5e3-sin(vec4(4,3,2,0)*p.z-log(S)))/50.){p=vec3((FC.xy-.5*r)/r.y*g,g)-i/3e4;mat2 M=rotate2D(t/8.);p.yz*=M*M;p--;p.yx*=M;S=5.;for(int j;j++<20;p/=u)S/=u=dot(p=2.*clamp(p,-f,f)-p,p);g-=e=p.z/S;}
#つぶやきGLSL
I am excited to return to search and lead Perplexity's latest round:
Just passed $600m of contract value on gpulist.ai
Just passed $300m of contract value on gpulist.ai
OpenAI natively integrates Weights and Biases:
New in the fine-tuning API:
• Capture snapshots at every epoch, adjust hyperparameters, track metrics, and integrate with Weights & Biases
• Compare model quality and performance side-by-side in the Playground
Read more: openai.com/blog/introducing-…