Thta DeepSeekMath-V2 scoring 118/120 on Putnam is absolutley wild, especially when human participants only hit 90. Makes me wonder if we're actualy approaching genuine symbolic reasoning or if these models are just finding pattterns we dont see. Either way, love how you packaged all these updates without the usual hype.
Thta DeepSeekMath-V2 scoring 118/120 on Putnam is absolutley wild, especially when human participants only hit 90. Makes me wonder if we're actualy approaching genuine symbolic reasoning or if these models are just finding pattterns we dont see. Either way, love how you packaged all these updates without the usual hype.