Keep on to blur preview images; turn off to show them clearly
probably an inference issue for Kimi. But I'm reminded again that DeepSeek has improved its post-training stack a lot in the last year. V3.2 really has no degradation from V3.1, and is way better than both R1s. Best open model on this test.
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
now i need to spend some more time and attune myself to the writing ability or lack of for this new model. also calibrate how I feel about it's intelligence inc or decrease.
RL and distributed training • eXperiments lab
RT @maxtempers: In the UK, we have de-automated smoke detectors. Buildings across Britain now have roving 24/7 'fire patrols' conducted by…
AI Optimist. Empiricist, not 'rationalist'. Anti world government.
我觉得从内容生成角度上看，世界模型有两种不同的意义 这也是23年和 @fuxiangPro @vista8 一起聊到AI温度钟摆的两端：极度严谨还是创造想象 - 从创造想象而言：世界模型能把人们带到更多拟想的平行世界，比如复原古老消失的建筑群，穿越回曾经的过往 - 从极度严谨上看：世界模型可以复制现实世界，带人们重新观测受限于经济，时空，无法到达的地方，去感知不一样的世界 我喜欢世界模型有两个原因，以上是一个超时空的原因，还有一个原因是世界模型能帮助AI在时空上有更强的数据反馈，这是通向现实AGI的一条大道 从前LLM受限于文本等模态，这些模态都有较多的信息损失，但世界模型弥补了这些 我选择一张我曾经看过的世界，让模型去生成，哪怕当下还受限于精度，但看起来已经相当震撼了 震撼的原因并不是因为它清晰，而是因为走向了不一样的远方
Believing is seeing
RT @DGMD22: Had a great time testing out Otherside Nexus @OthersideMeta. Some thoughts! Bottom line is the engine and world feels good, t…
Investor & Collector.
Just shipped Chartbrew AI to production! 🚀 The AI assistant can be launched from anywhere in the app to help gather insights, create datasets, and add charts to reports Lots to add and improve, but hopefully I can gather some feedback to go in the right direction 🤞
Available now at https://t.co/OYrOyKlT9O ✌️
