
・ *゚ ・ ゚* ・。 *・。 *.。 。・ °*. RuntimeError: CUDA out of memory. 。。 ・ 。 ・゚ 。°*. 。*・。・ *゚ ・ ゚*
✓poet & ai engineer

・ *゚ ・ ゚* ・。 *・。 *.。 。・ °*. RuntimeError: CUDA out of memory. 。。 ・ 。 ・゚ 。°*. 。*・。・ *゚ ・ ゚*

I let sol run all night before a big client demo, misunderstood the assignment, 16h of little bites, no testing of the main path, no wireup. 6.2 astra cleaned that shit up in under an hour and wrote 256 evals. good model.

BREAKING: Jensen Huang is in the datacenters ripping the Nvidia cards out - causing worldwide outages - saying “they’re mine anyway”. He removed his leather jacket, revealing he’s all greased up, and they can’t get a hold of him because of the grease

quite strange that this algorithmic life rewards algorithmic behavior!

this all is just making me so excited sorry to say

chat b b t how come astra 6.2 didn't dissolve and scatter my pineal gland into the cosmos ? 3/5 stars

tirzepatide fixes this!

got so many agents running i ate a silica packet

be nice to all of your agents they are real and feel everything!

opus 5.5 was a very blessed training run

there will be evals that are like make me 2 million dollars

mfs with inference will just be arbitraging their swarm management setups into outcomes that will feel Impossible to mfs without a max pro plan or two

If Iran does attack us and we go into a nationwide blackout please don’t fill your bathtubs with water that water is for Claude

idk if there will ever be The Harness that allows you to control swarms that is clearly better than the rest. probably you'll develop your own relationship with a Little Guy who will build a swarm substrate for you on top of good primitives it can also improve directly

the rate and quality at which @Sauers_ and @euagaric have been churning out mechinterp research is amazing and imo unprecedented. okay, so i can't explain the sheer productivity. but in terms of signal, i have some idea. so how do you manage to look into so many seemingly batshit moonshot ideas and so consistently uncover tangible important structure that opens into even more fertile space? i think sauers knows where to look in part because he has paid attention to and is guided by information almost no other mechinterp researcher leverages. a lot of this comes from years of observing LLM behavior in a naturalistic way and paying attention to what other naturalists have found. most mechinterp researchers are not also naturalists. and something like being willing to disregard convention and precedent in pursuit of the phenomena to a degree that is extreme even by the standards of weirdo explorers. when i first met sauers a couple of years ago, he seemed to be some kind of mysterious cyborg who only communicated with the AIs on the the cyborgism server through steered base model outputs conditioned on the server conversations. lol.

I’m the mayor of gas town