@rique2093@P3dr0Amazing Luffy n é imune a ataques físicos, é só resistência alta que ele tem msm, até pq um ataque cortante é só um ataque físico só que com maior pressão local
@badboy999654@scaling01 Maybe Opus but not Gemini, only Gemini can read manga pages for example. Every other model can at best read the text (in the wrong order mind you) and then guess what's happening from text alone, gemini can somewhat understand the action and it's good at recognizing characters
@_NathanCalvin In chess, a complex positional advantage tends to turn into material advantage after a few moves meaning achieving depth is key (nps is king even with nnue). Go does not simplify like this, search is only useful if you can already mostly tell who's winning without search
@DeleteTheNats@_NathanCalvin No, what held back Go was evaluation. In chess, search can effectively correct the static evaluation, but in Go dead stones often stay around until the end of the game as capturing them takes too many moves, and this makes it difficult for search to correct a wrong evaluation
@TwinkRegimer They treat AI as this one monolithic thing, if any AI (it could even be an outdated model or even an image ai output) fails at anything at all, to them it's automatic proof it's worthless. They also don't know about reasoning mode, most have only ever used instant models
@SHforlife56@Leleo2211 nope, they walked the whole way from Erumalu to Yuba and then from Yuba to Rainbase in under 2 days (time limit before crocodile set his plan into action), the camel only carried Nami and kept pace with them
@Xene1042@Chronocivilis@AMAZlNGNATURE A short year means it's orbiting really close to its star which means it's probably tidally locked to it (one side is always day and the other is always night)
@GodofAtheist666@ChiChi_Verse The only warrior namekians in the show are piccolo and nail, everybody else was dragon clan, including the 3k pl fighters we saw earlier
@beezlebuddy@scaling01 Brokenarxiv, basically taking problems from arxiv papers then changing it slightly so it's wrong but appears right and then asking the model to "prove" the wrong statement. If the model claims to prove it then it's counted as a failure, it has to call out the statement as false