oforecu

LLM들한테 복잡한 업무 시키면 내가 뭐 말했는지 문맥도 파악 못하고 "아 맞다, 그 얘기였죠?"라며 헛소리하거나 문맥 다 까먹어서 처음부터 다시 설명해야 하죠. 이 이유는 LLM들의 기존 메모리 시스템 정확도가 고작 0.2141이라니이기 때문인데 @OpenGradient 의 MemSync는 단순한 데이터 저장이 아니라, 여러 대화에 흩어진 문맥을 연결하는 Multi-hop의 'Universal Memory Layer'를 통해 진짜 사람처럼 기억하게 만들어 줍니다! 또한 @OpenGradient 는 243% 성능 향상이라는 시스템 정확도 수치를 가지고 있다고 합니다! 벤치마크 결과 0.7344 (MemSync) vs 0.2141 (기존)... 맨날 치매 걸린 AI만 보다가 AI가 똑똑해지면 좀 이상할 것 같지만 그래도 진짜 보급화가 필요하다 봅니다! 사랑해 @OpenGradient!

Some big personal news… I’m joining the @opensea team as a Marketing Lead. I get to wake up everyday and work on things I’m passionate about…leading Communications, driving product and marketing for Rewards/Loyalty, and guiding social strategy. Community. Storytelling. Product. It’s like a bingo card of the things I love to work on. Doing that at any company would be exciting, but to do it at OpenSea, one of the most impactful companies in crypto, is the kind of opportunity you dream about. This also means I’ll be stepping away from my role as cohost of Coffee with Captain. CwC has been one of the best experiences of my life, and I’ll have more reflective thoughts on that unforgettable chapter in the coming days. But for today, it’s all about the next chapter…one I’m beyond excited to begin writing. OpenSea played a huge role in shaping my Web3 journey. Now I get to help shape its future. Sails up. ⛵️