DerekZ.KAS

@WhaleChart Buy $GROK. Never stop.

TRENDING MEMECOINS 🚀 $BAM 0x9db0fb0aebe6a925b7838d16e3993a3976a64aab $MONG 0x1ce270557c1f68cfb577b856766310bf8b47fd9c $DINO 0x49642110b712c1fd7261bc074105e9e44676c68f $LIZA 0x8b227d72570d3ead66014bca8305cbef7f90d1ee $GROK 0x8390a1da07e376ef7add4be859ba74fb83aa02d5 What are you buying today?

What's @RealCharAI tech stack? v0.0.3 LLM related - Text to Speech: @elevenlabs,@googlecloud,@UnrealSpeech - Speech to Text: Whisper,@googlecloud - VectorDB: @trychroma - LLM: @OpenAI,@AnthropicAI,@MetaAI (LlamaV2) - LLM Framework: @langchain - LLM Data: @llama_index 🧵👇

一个大新闻：Facebook 开源了一个专注于声音的 Generative AI 工具： AudioCraft ai.meta.com/blog/audiocraf… 输入文字描述，就可以生成音频，我试了一下，确实很强大。 分为三部分： MusicGen：主打音乐生成 AudioGen：主打各种音效生成 EnCodec： 智能音频压缩。 AudioGen 其实是更强的，这里有一些 demo，惊为天人，felixkreuk.github.io/audiogen/ MusicGen，现在在 huggingface 上可以随便玩，huggingface.co/spaces/faceboo… 大家可以感受一下。 Github 地址在这里：github.com/facebookresear… 已经 9.8k 个 star 了。

arxiv.org/abs/2307.16892 #lk99

超导看来真的要来了。=> 费米能级的孤立平带是超导晶体的标志。

内容创作者每月可以赚取超过10,000美元 ChatGPT 和 AI 工具使得这一切变得更加简单 以下是 6 款可以帮助你在网上“印钱”的AI工具： 1/ 内容生成： writesonic.com 2/ 针对网络的研究操作系统： nette.io 3/ 生成病毒式推文和话题串： postwise.ai 4/ YouTube视频编辑器： FlexClip.com 5/ 文本转视频： synthesia.io 6/ 设计工具： designer.microsoft.com

No doubt, ChatGPT is awesome. But the next big thing in AI could be, GPT-engineer – Write a prompt about what you want to build – AI will ask for clarification – and then generate the codebase github.com/AntonOsika/gpt…

google的DeepMind推出全新AI机器人代理RoboCat，一款能自我提升、操作多种机器人的神器！ 它能快速的在不同机械臂上学习并解决新任务，更能自我生成训练数据以自我提升。 这个AI机器人领域的突破有多重要？对全能机器人出现将起什么作用？技术上如何实现？ 🧵

It only took me 2 DAYS to create this beautiful Lofi animation 🔥🔊 Bringing AI pictures to life is way easier than you think. EVERYONE can do it. These are the 8 steps you should follow 👇

So...I thought Photoshop was in trouble because of AI. But now, Adobe's new AI-enhanced Photoshop beta is out. And it's one of the most magical things I've ever seen. (part 2) 12 game-changing examples that'll blow your mind:

Excited to announce our multi-modal agent API that can automatically chain 100,000+ of HF models (stable diffusion, whisper, OpenAssistant,...) for text, audio, image, video, time-series,… based on your commands. All open-source so it can work locally! huggingface.co/docs/transform…

We just released Transformers' boldest feature: Transformers Agents. This removes the barrier of entry to machine learning Control 100,000+ HF models by talking to Transformers and Diffusers Fully multimodal agent: text, images, video, audio, docs...🌎 huggingface.co/docs/transform…

Bitcoin has come a long way since its inception in 2008. It has evolved from a mere idea of a decentralized digital currency to a full-fledged network that has the potential to change the way we transact and trade globally. However, the network is facing a major challenge - scalability. The current transaction processing limit of 7 transactions per second is not sufficient for Bitcoin to become a mainstream currency. The solution is to scale on layer 1, which means increasing the block size limit and making it possible to process more transactions per second. Scalability is not just about improving transaction speed, but it also has implications for transparency in trade globally. On-chain data can provide a high level of transparency that is not available in traditional financial systems and corrupt governments. The immutable nature of the blockchain allows for a transparent record of all transactions, which can help to prevent fraud, corruption, and money laundering. In addition, on-chain data ownership is a crucial aspect of the Bitcoin network. It allows individuals to have control over their data and decide how it is used. This is in contrast to traditional financial systems where data ownership is often controlled by centralized institutions. With Bitcoin, individuals can choose to monetize their data. The monetization of data is a potential new revenue stream for individuals and businesses. By having control over their data, individuals can choose to sell it to their fans and 3rd parties, which can result in a new source of income. This can be especially beneficial for those in developing countries where traditional financial systems are not accessible. The need for Bitcoin to scale on layer 1 is vital for its continued success. On-chain data creates transparency in trade globally, and data ownership and monetizable data are key components of the Bitcoin network. By addressing the scalability issue, Bitcoin can become a mainstream currency that is accessible to all, and its potential for changing the way we transact and trade globally can be fully realized.

分享一个强大的开发人员分享快速参考备忘清单(速查表)，可以快速查找和使用各种技术和工具，包括前端、后端、数据库、操作系统等等，目的是帮助开发人员技术栈查阅，提高效率和减少出错率。 项目：github.com/jaywcjlove/ref… 镜像： quickref.cn

IMAGEBIND: One Embedding Space To Bind Them All. Learns a joint embedding across six different modalities - images, text, audio, depth, thermal, and IMU data. An open source project by Meta-FAIR. Paper: dl.fbaipublicfiles.com/imagebind/imag… Demo: imagebind.metademolab.com Code: github.com/facebookresear… Blog post: ai.facebook.com/blog/imagebind…

Meta 最近常常放大招，今天发布了新的开源多模态模型 ImageBind，可以将文本、音频、视觉、温度还有运动数据流串联在一起，形成一个单一的 embedding space，可以让机器能从多维度来理解世界，也能创造沉浸式的多感官体验，Metaverse 需要 AI🧵ai.facebook.com/blog/imagebind…

If you look under the hood kernels play a big role in many areas of ML From Attention, to MLPs (including NERF) and more, they're explainable w/ kernels that adapt to data values and are encoded w/position Here' a very nice overview aimed at non-experts bit.ly/3xehWM6

Introducing InfiniteGPT ♾️ — a Python script that lets you use the chatGPT API with an unlimited amount of text. ❌ No more re-prompting 😖 No more tedious copy & pasting into chatGPT Drop a message below to get the link!

ZipIt! Merging Models from Different Tasks without Training ai.papers.bar/paper/0667a582… Deep visual recognition models are capable of performing the one task they were trained on. We find that this fails to account for... 🧵 👇

整个数据科学领域即将被革命化。#ChatGPT #CodeInterpreter 这里有 10 个疯狂的例子来证明这一点：