3 ms·
Traders are saying not doing multitoken prediction, not using Sharpe ratio adjusted rewards, using reward models, and not compressing KV cache tokens by >90%, w
by pizza 2y ago
Traders are saying not doing multitoken prediction, not using Sharpe ratio adjusted rewards, using reward models, and not compressing KV cache tokens by >90%, were supposed to be worth hundreds of billions of dollars of future expected revenue flow, at least according to other traders.
I say to the traders: you should have just stuck to reading arxiv, TPOT, and jhana twitter for the past 2 years, rather than listening to other traders, if you were trying to understand the utter spread of low hanging fruit that just hasn’t been picked up yet!
- igleria 2y ago> you should have just stuck to reading arxiv, TPOT, and jhana twitter for the past 2 years Traders expect to be able to trade based on regurgitated headlines made by tech influencers who don't know ANYTHING about NOTHING.
- djtango 2y agoWho is jhana?
- pizza 2y agoIt’s a part of twitter that is nominally about advanced forms of meditation but is also a surprisingly overrepresented by ai researchers lol
- nejsjsjsbsb 2y agoYou need to be very solvent to stay rational.
- danielmarkbruce 2y agoThe low hanging fruit thing is 100% correct. Anyone reading papers saw it everywhere, on every dimension. And it's not to say the authors didn't see it either, they did - they just had to get something out now. I'd guess folks in semi conductors saw the same things for ages.