How ChatGPT was really made - Lex Fridman

#AI #Training #Models

Original timestamp: 00:44:52

AA (@measure_plan)

작성자는 피아노 스케일 연습과 디스코드에서의 하이스코어 경쟁 때문에 'opus 4.6'과 'gpt 5.3' 출시(런칭) 열풍을 놓쳤다고 밝힘. 트윗은 두 모델의 최근 출시 소식이 화제가 되었음을 암시함.

https://x.com/measure_plan/status/2019537094700679300

#gpt #gpt5.3 #opus #ai #models

Two new .release for frontier #models just dropped.

#Claude 4.6 Opus and #ChatGpt 5.3

#AI #LLM

Bindu Reddy (@bindureddy)

모델/버전 추천 목록: Kling 3.0(비디오) 출시를 알리며 최종 모델 업데이트를 제시. 코딩: Opus 4.5, 베스트 스몰 모델: Gemini Flash 3.0, 비디오: Kling 3.0, 이미지: Nano Banana Pro, 파인튜닝 베이스: GLM 또는 Qwen, OCR: Flash 2.5, 데이터 분석: GPT 5.2(생각 중).

https://x.com/bindureddy/status/2019196229121638485

#models #benchmark #kling #gpt

“But current #economic #models used by #governments and #financialinstitutions entirely miss such shocks, the researchers said, instead forecasting that steady #economic #growth will be slowed only by gradually rising average #temperatures” www.theguardian.com/environment/...

Flawed economic models mean cl...

Only just recently had some time to read the Kimi K2.5 technical report , an impressive open weights model (it's not fully opensource), the distance towards large paid proprietary foundation models is becoming very small.
https://github.com/MoonshotAI/Kimi-K2.5/blob/master/tech_report.pdf
It also feels that way in use:
https://www.kimi.com/chat
#AI #openweights #models #KimiK25 #China

Low Poly Cyber City Buildings #Lowpoly #Building #Models #Textures #Gamedevelopment #Simulation #Survivalgames #AssetStore

https://u3dn.com/packages/low-poly-cyber-city-buildings-242424