The fact that we as a society are still capable of producing art this beautiful gives me actual hope for the future. The fascists can never do something like this.
In retrospect, I should have waited to buy @OpenAI credits 3 weeks ago. Who knew @deepseek_ai would perform comparatively well, at a fraction of the cost!
NEW: Danish officials are "utterly freaked out" & in "crisis mode" after Trump told them he intends to acquire Greenland during a 45-min call.
Trump was firm in his pursuit to acquire Greenland during a call with Denmark's prime minister, according to the Financial Times.
Five European officials who were briefed about the call were in shock to find that Trump is serious about acquiring Greenland.
The officials hoped he was joking, or his statements were just a negotiating tactic.
"[Trump] was very firm. It was a cold shower. Before, it was hard to take it seriously. But I do think it is serious and potentially very dangerous," one official reportedly said.
"The intent was very clear. They want it. The Danes are now in crisis mode. The Danes are utterly freaked out by this."
"It was a very tough conversation. He threatened specific measures against Denmark such as targeted tariffs."
i feel like the older you get, the more quiet you become. life humbles you so deeply as you age. you realize how much nonsense you've wasted time on. you start to accept things for what they really are. you stop forcing friendships & connections with people & you just learn to grow
Two reasons you don’t want to get carried away with way too many “skilled immigrants” are:
1. You shouldn’t displace your own young men
2. You actually want CULTURE FIT more than smarts. The balance of dynamic young men in your society should respect each other and get along;
DeepSeek (Chinese AI co) making it look easy today with an open weights release of a frontier-grade LLM trained on a joke of a budget (2048 GPUs for 2 months, $6M).
For reference, this level of capability is supposed to require clusters of closer to 16K GPUs, the ones being brought up today are more around 100K GPUs. E.g. Llama 3 405B used 30.8M GPU-hours, while DeepSeek-V3 looks to be a stronger model at only 2.8M GPU-hours (~11X less compute). If the model also passes vibe checks (e.g. LLM arena rankings are ongoing, my few quick tests went well so far) it will be a highly impressive display of research and engineering under resource constraints.
Does this mean you don't need large GPU clusters for frontier LLMs? No but you have to ensure that you're not wasteful with what you have, and this looks like a nice demonstration that there's still a lot to get through with both data and algorithms.
Very nice & detailed tech report too, reading through.