Keep on to blur preview images; turn off to show them clearly
RT @doodlestein: @mathelirium Yup. It’s also why cosine similarity works so well in high-dimensional vector spaces (like embeddings in LLMs…
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM
I developed something extremely similar in Unity for The Last Night over the years. Crucial toolkit to depict crazy AI architecture based on biomimicry & kinetic principles, designed like an interactive theme park to delight humans.
Founder @oddtalesgames Directing The Last Night @TLN_Game Art Direction, Cinematography, Tech Art. Atoms, Bits, Memes, Genes. Freedom, Futurism, Humanism.
also my motto in my childhood It means: - self-reliant and never stop improving. - bear everything with great virtue and tolerance.
Staff @Kimi_Moonshot prev. co-maker of ModelizeAI & gemsouls "Personality goes a long way" @UCSanDiego
The Neocognitron was so far ahead of its time. Fukushima doesn’t get enough credit. Great name, too. I hope he’s famous in Japan at least.
Former Quant Investor, now building @lumera (formerly called Pastel Network) | My Open Source Projects: https://t.co/9qbOCDlaqM
爱来自法国, Mistral 3 Large 模型发布啦! 不过貌似赶了个晚集? 这次的 Mistral 3 Large 终于更新为了 MoE 架构，总参数量 675B, 激活参数量41B, 上下文长度 256K. 他们自己测试，基座模型性能上超过了 DeepSeek-V3.1 不过我还是略表怀疑，因为 Mistral 系列在我之前测试都没有什么深刻的印象。 另外本次还一并发布了 Ministral 14B, 8B, 3B. 官方也是直接抓着 Qwen3 的同等参数量大小的模型对比了一番. 如果感兴趣的同学多我会考虑给大家做简单测试.
模型参数/1
RT @grsl_fr: I'm really happy about the first comment. But I'm even happier about the response. People now think that AI-generated actors…
I just launched a new startup today → VIDEO AI ME → Create your own AI clone and make studio-quality videos that go viral → https://t.co/eU2hD1uyoV