“You want maximum utilization in the cloud if you’re trying to deal with this very supply-constrained environment. And somehow, many people’s intuition is the opposite of that,” Altman
oh yesss, RAM producers gonna love this :D Jokes aside very interesting experiment. Giving a fully functional cpu ram and hd, so they can technically hide stuff better and avoid getting caught in COT (yikes).
No practical implications, we dont use 10^87 variables (a point where the proposed new method surpasses the already existing sat solvers), however a good theoritical point.
People need to think about the future now, it is more urgent than before, we have to decide what to do with our life when (assuming a positive future rather than skynet) AI does the stuff better than us in every category. We have to find a meaning. I know this comment isn't directly related to the AI trades, but every subject regarding AI keeps reminding me of the future we are not ready. I can't imagine what type of professions completely made unnecessary by AI in the following 5 years.
That's matter cruelty. We need more ethical ways to study those poor things. Does anyone know what happened after the experiment ? Did they finally let it go?
Exagerating? I setup llm endpoints on corporate level in a company on H200 gpus, that serves hundreds of software engineers, I have observed differences, even on different versions of nvidia drivers. I shared my day to day experience. It is more and more important if you have very limited VRAM like 24gb on a 4090 RTX desktop or RTX 5090 Laptop gpu. I set up and use llms on my personal devices also, when I can't use cloud models on classified company materials. Here is an example problem what I daily report and deal with.
Model list is a bit limited, and whole harness affects the speed, so the libs used in harness and their versions might be useful to show for each individual model setup.
Flying cars, Super Duper Intelligence, AGI and yet we can't even stop these, we can't even have peace on earth, we can't avoid conflicts. What good these technological advances are if we can't prevent ourselves being violent apes?
I dont think Opus 4.8 is an average coder, with my own experience (I have coded 20 + years before even llms existed) it is anything but average. I don't think training data alone determines the success of these models, there are lots of reinforncement learning principles and fine tuning takes place, a crappy code in the dataset doesnt hold those llms scoring high in benchmarks, I dont think an average programmer can score 70% (opus 4.8) in SWE Bench Pro, which is a good one.
I don’t think anyone can predict 1 year later. It is all in the hands of few select people. After today’s news -fable being suspended- I am not sure what to think of future jobs. I can only guess that human connection will be more important in the future.