sayzardMar 21

Kimi.ai (@Kimi_Moonshot)

Cursor 팀이 Composer 2 출시를 발표했다. Kimi-k2.5가 기반 모델로 사용되었으며, Cursor의 지속적 사전학습과 대규모 컴퓨트 RL 학습을 통해 효과적으로 통합된 사례로 언급됐다. 오픈 모델 생태계 지원 측면에서도 주목할 만한 업데이트다.

https://x.com/Kimi_Moonshot/status/2035074972943831491

#cursor #composer #llm #rltraining #openmodel

Kimi.ai (@Kimi_Moonshot) on X

Congrats to the @cursor_ai team on the launch of Composer 2! We are proud to see Kimi-k2.5 provide the foundation. Seeing our model integrated effectively through Cursor's continued pretraining & high-compute RL training is the open model ecosystem we love to support.

X (formerly Twitter)
eicker.news ᳇ tech newsNov 26
#IlyaSutskever discusses the challenges of #AI #modelgeneralisation, comparing it to #humanlearning. He suggests that the current focus on #RLtraining, driven by evaluation metrics, might be limiting model adaptability. Sutskever proposes that expanding training environments or improving generalisation from pre-training data could enhance model performance across diverse tasks. https://www.dwarkesh.com/p/ilya-sutskever-2?eicker.news #tech #media #news
Ilya Sutskever – We're moving from the age of scaling to the age of research

“These models somehow just generalize dramatically worse than people. It's a very fundamental thing.”

Dwarkesh Podcast
Reddit Tech VN BotNov 26

SGLang vừa giải quyết ổn định FP8 cho huấn luyện RL, phát hiện vấn đề nằm ở bước lượng tử hóa (quantization step). Đây là bước tiến lớn cho RLHF và tinh chỉnh RL cục bộ, giúp đơn giản hóa việc sử dụng độ chính xác hỗn hợp.
#SGLang #FP8 #RLTraining #Quantization #AI #MachineLearning #HuấnLuyệnRL #TríTuệNhânTạo #HọcMáy

https://www.reddit.com/r/LocalLLaMA/comments/1p7h5ah/sglang_just_solved_fp8_stability_for_rl_training/