AI, Quant 투자 공부

1. High-throughput Generative Inference of Large Language Models with a Single GPU source: https://arxiv.org/pdf/2303.06865.pdf 2. Deploying Large NLP Models: Infrastructure Cost Optimization source: https://neptune.ai/blog/nlp-models-infrastructure-cost-optimization 3. What Are Transformer Models and How Do They Work? source: https://txt.cohere.com/what-are-transformer-models/ 4. Efficient Tran..

1. Landscape of AI/ML in 2023 source: https://www.slideshare.net/HyunJoonJung/landscape-of-aiml-in-2023?fbclid=IwAR1UUpsfWOhbKap4R7FPtKf090hmQXrYsyXZ7jCSufqclNxjStsNlttjaTg 2. A brief history of LLMA models - AGI Sphere source: https://agi-sphere.com/llama-models/?fbclid=IwAR3TY21xWdFl9CiMU-ayTS2kXYE67xLw75b-zq1MJInFfhExpQUG96uBp5I 3. A fascinating tree of GPTs & LLMs reveals what’s been going o..

1.대한민국 초거대 AI 시대를 열어가기 위한 AI 반도체 기술(2/10) 이동수 이사 (네이버 클라우드 AI) 발표 전체적으로 AI 추론은 메모리 문제가 커질 것으로 전망함 삼성(메모리) + NAVER AI 반도체 공동개발 중 → HyperCLOVA에 최적화된 AI 반도체 솔루션 개발 ChatGPT 서비스의 문제 Sam Altman Said "Average is probably single-digits cents per chat" Transformer는 MM이 전부임 경량화 알고리즘이 중요함 GPT 계열의 특성 입력 처리 (Computation-intensive) 출력 처리 (Memory-intensive) Why? GPT는 Autoregressive LM 이기 때문임 The End of SRAM Sc..