markets arereallyus
@tao_te_chaching
Followers
592
Following
59K
Statuses
4K
and where does the newborn go from here?
New York
Joined January 2021
I'm an indie guy who wears athletic clothes (bike 20-40 miles every week), has a degree in psychology, somehow now works at several AI companies, and plays instruments. Don't judge us all the same ✌️
0
0
3
neo-foidalism. is that anything. or techno-foidalism. e-girls are simply foids that have been reterritorialized under subprime attention parasocial incel capitalism, aka techno-foids. is this anything.
0
1
11
the great thing about the politically polarized umich inflation expectations poll is that we can just check to see which side was more accurate a year from now
0
0
0
RT @tao_te_chaching: everytime a new version of chatGPT comes out theres a guy who goes "you don't get it, i get so much value out of this.…
0
1
0
finding out that the CFTC regulation ("section 40.11") that prohibits assassination event contracts (ie, assassination markets) specifically prohibits "terrorism, assassination, war" and, finally, "gaming"
0
0
0
she let me hit because of my unified data-driven ontology
0
0
3
personally i thought we already mostly got rid of AIDs in the US. come to find out we've actually been funding it?
0
1
1
@HannesThurnherr @suchenzang @tylerlu @tdietterich i'm not sure, but some sort of synthesis of these results is required. "does a longer thought trace improve the final answer?" wasnt meant as a rhetorical question.
1
0
3
@HannesThurnherr @suchenzang @tylerlu @tdietterich the comparison between the correct solutions (green) bar and incorrect solutions (red) bar are for the same tasks (therefore accounting for task difficulty). the post says "Even on the same question" and details a method by which the same prompt (aka task) is run multiple times
2
0
7
Discovered a very interesting thing about DeepSeek-R1 and all reasoning models: The wrong answers are much longer while the correct answers are much shorter. Even on the same question, when we re-run the model, it sometimes produces a short (usually correct) answer or a wrong verbose one. Based on this, I'd like to propose a simple idea called Laconic decoding: Run the model 5 times (in parallel) and pick the answer with the smallest number of tokens. Our preliminary results show that this decoding gives +6-7% on AIME24 with only a few parallel runs. I think this is better (and faster) than consensus decoding.
1
0
14
sometimes, during particularly heated training sessions, she ends up striking me a couple times before regaining her composure. she's always apologetic and i act hurt but i secretly get off on her beating me. nowadays, if want an erection i can simply snap my fingers
0
0
2
monday morning weekend sentiment check note that tallies the number of "it's over" chud posts vs the number of "nothing ever happens" chud posts
0
0
0
miss cottrill... i'm not like these other guys... if you'd just get to know me a little .... miss cottrill...
0
0
1
2 weeks ago trumpcoin+melaniacoin were 4D chess moves to cause deflation via "liquidity soaking". today tariffs are 4D chess move aimed at timing an inevitable market rebound just in time for midterm elections. where my 2nd and 3rd order thinkers at?
Trump has every possible incentive to tank markets right here and now. 1. Nothing will kill inflation faster than sustained asset price deflation 2. He gets rate cuts if equities and credit tighten FCI 3. Get all the policy tightening - border, trade, fiscal - out of the way now, then rip markets for the rest of his term on the back of low rates/inflation, AI/re-industrialization boom, tax/reg cuts. “Turn the bull on” July 4, 2026.
0
0
1
corollary question: how much larger would a female dog need to be for smaller male dogs to stop recognizing it as a potential mate? the smaller dogs here are leghumping a dog 5x their size. would their instincts have them toehumping a dog 50x their size?
0
0
1