Keep on to blur preview images; turn off to show them clearly
Well, Mistral has definitely reemerged with a certain Whale flavor But seriously, this seems to be about some generic civilizational will-to-live. Why is there no Arcee, PI, Zyphra in the EU or Japan? It's dozens of people. You're not *that* brain-drained nor compute-poor.
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
Well, Mistral has definitely reemerged with a certain Whale flavor But seriously, this seems to be about some generic civilizational will-to-live. Why is there no Arcee, PI, Zyphra in the EU or Japan? It's dozens of people. You're not *that* brain-drained nor compute-poor.
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
开源分解图层大模型这么快就来了? 给大家带来阿里刚刚发布的 Qwen-Image-Layered 模型实测, 这是一个可以把图片分成不同图层的大模型, 模型是基于 Qwen-Image 微调的. 本次我的测试覆盖这个模型的擅长场景(海报), 指令遵循测试(指定抽取目标), 边缘处理(头发), 极限测试(全是贴纸, 一个贴纸一个图层的话可以抽取50多个图层). 直接说结论, 首先还是模型太大了, 因为这个模型基于 Qwen-Image 所以是个20B的模型, 我是用HuggingFace Zero GPU 测试, 每次运行大概2分钟左右, 模型的确可以做到分离图层, 而且边缘处理很不错, 不过稳定性有待优化, 在我测试中4个图层可以输出, 但是8个或者10个图层就炸了, 怀疑可能是超过了Zero GPU超时或者bug (GPU是H200, 不太可能爆显存), 输出尺寸只有 544*736, 官方也推荐640分辨力, 这点也需要提升, 以及模型还是太大了, 20B, 希望能优化下尺寸.
A coder, road bike rider, server fortune teller, electronic waste collector, co-founder of KCORES, ex-director at IllaSoft, KingsoftOffice, Juejin.
> It also means updating our export control laws to recognize that the CCP has used Chinese national researchers as tools of their tech theft strategy yes. I wonder what form this recognition will take. Deportation? Well… Realistically they need to do a Qian Xuesen again.
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
Not really decisive on a ship, I think. Even 100 KW needs seconds of lock on a serious drone. You can saturate this, and against missiles it's pretty hopeless. Scale this up another OOM, and we have something interesting.
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
RT @iAnonymous3000: Read through Coinbase Wallet (now Base App) privacy policy. The "self-custodial" branding obscures significant data col…
Privacy Experience @PrivacyEthereum | Events @ethereum | OSS 🛠️ Indie Maker @eth_gastracker https://t.co/xNlTAwRRp1