정지훈 Jihoon Jeong

@hiconcep

Startup Investor. Artificial Intelligence, Metaverse & Blockchain Tech Evangelist. General Partner @ K2G Tech Fund, Advisor of LG Electronics and Lunit

37.518042,127.041889linkedin.com/in/jihoonjeong/Joined December 2008

1,022 Following

127.7K Followers

정지훈 Jihoon Jeong 🗣’s Tweets

Google announces Dreamix: a model that generates videos when given: - video + prompt (Video editing) - input images + prompt (Subject Driven Generation) - input image + prompt (Image-toVideo

0:48

424.3K views

1,949

8,431

Tired: train a large language model (LLM) to generate text in human languages Wired: train an LLM to generate proteins sequences, the language of life In this recent paper, researchers trained a decoder-style LLM to generate functional proteins: nature.com/articles/s4158

169

848

The biggest infrastructure provider in the world is targeting Web3 for growth. E-commerce giant

@Amazon

's cloud service,

@awscloud

, is hiring staff to help increase its clientele in the Web3 space.

coindesk.com

Amazon Web Services Hiring to Pump Its Web3 Clientele

The biggest infrastructure provider in the world is targeting Web 3 for growth.

227

정지훈 Jihoon Jeong

@hiconcep

ChatGPT (2) : ChatGPT를 보통 GPT와 구분시킨 RLHF

youtube.com

ChatGPT (2) : ChatGPT를 보통 GPT와 구분시킨 RLHF

지난 시간 올린 ChatGPT에 대한 기술 분석 이야기가 반응이 좋아서, 조금 더 부지런히 제작을 해서 업로드를 하고자 합니다. 이번에는 ChatGPT를 일반적인 GPT 대비 이렇게 서비스 품질을 좋게 만든 핵심기술이라고 할 수 있는 RLHF (Reinforcement Learni...

This is a mathematics joke. Source: bit.ly/3wUOPPa

1,132

6,957

A tutorial from

on how to build a question-answering system on your custom website Preprocess: 1. Website -> text (Scraping) 2. Text -> chunks -> embed Inference: 1. Question -> embed 2. Get topK similar chunks 3. Answer generation using 1 & 2 platform.openai.com/docs/tutorials

191

1,016

what level is this

0:28

332.5K views

110

1,577

5,644

정지훈 Jihoon Jeong

@hiconcep

Feb 4

ChatGPT (1) : GPT는 뭐고, ChatGPT는 또 뭔가?

youtube.com

ChatGPT (1) : GPT는 뭐고, ChatGPT는 또 뭔가?

거의 1년 6개월 만에 동영상 업로드합니다. 그렇게 오랜동안 업데이트를 하지 않았음에도 놀랍게도 제가 업로드 열심히 할 때보다 방문자가 늘었더군요. 분석을 해보니 딥러닝을 포함해서 AI에 대해 알고 싶어 하는 분들이 꾸준히 찾아 주시길래, 그런 분들에게 조금이라도 도움이 되고자 짬...

These synthesized video results are surprisingly realistic and genuinely fun to watch. scenescape.github.io

0:05

7.2K views

101

Synthetic life forms in their digital domains. Artificial Organisms, 2021.

0:11

9.4K views

670

The developer behind Cosmic Trip and Fujii,

@funktroniclabs

is back with The Light Brigade—a roguelike VR shooter that combines realistic, physics-based mechanics with powerful light magic. Pre-order now to save 10% off the purchase price. meta.com/blog/quest/lig

Today, we are opening up public access to a new AI product we have been building called Poe. Poe lets people ask questions, get instant answers, and have back-and-forth conversations with several AI-powered bots. (1/n)

398

1,671

Today's announcement with

@AnthropicAI

is another great example of the ecosystem of AI startups leveraging Google Cloud's reliable and open infrastructure to build their businesses. goo.gle/3JIAWem

116

728

Announcing our open source collaboration with

@huggingface

! 🚀 75+ #PaddleNLP models already on the

@huggingface

Hub 🔥 More awesome #PaddlePaddle models across text, image, audio, video and multi-modalities in the works! Learn more at

huggingface.co

Welcome PaddlePaddle to the Hugging Face Hub

174

To be fair,

had a big impact on AI/ML. Their OpenAI Gym env for RL, their CLIP - a de facto model for any image-text research, their Whisper speech recognition is amazing - all open sourced. In my view, they've had a bigger impact than many labs publishing 100s papers 🧵

Quote Tweet

Yann LeCun

@ylecun

Feb 2

Data on the intellectual contribution to AI from various research organizations. Some of organizations publish knowledge and open-source code for the entire world to use. Others just consume it. twitter.com/johnjnay/statu…

870

Facebook releases a 30B param “OPT+IML” (Open Pre Trained + Instruction Meta Learned) model fine tuned on 2000 tasks. Available on huggingface.

huggingface.co

facebook/opt-iml-30b · Hugging Face

111

653

Nature.

0:52

477K views

142

PADL: Language-Directed Physics-Based Character Control abs: arxiv.org/abs/2301.13868 project page: nv-tlabs.github.io/PADL/

2:18

26.5K views

244

825

OpenAI just released a new model to distinguish between AI/human written text to protect against ChatGPT. The classifier was trained on a pair of AI/human written dataset. However.. I was easily able to trick it by using GPT3 to rewrite the text. Demo: platform.openai.com/ai-text-classi

0:32

49.3K views

242

922

Animatronics were first introduced by Disney in 1962 for the film Mary Poppins (released in 1964). Since then they have come a long way. This is a giant T-Rex presented at BBC in July 2018 [source: buff.ly/2A96pEf]

0:19

1.2M views

From

Massimo

260

5,976

37.2K

Can LLMs understand images? We introduce 🔥BLIP-2🔥, a generic and efficient vision-language pre-training strategy that bootstraps from frozen❄️image encoders and frozen❄️LLMs. BLIP-2 outperforms existing SoTAs with only 188M trainable parameters! Github: github.com/salesforce/LAV

322

1,366

1/ Introducing high precision masked editing. No bleeding. Highly targeted AI based image editing to give you more control. Here's an end-to-end workflow in 30 seconds:

0:30

82.2K views

165

1,203

Excited to share SingSong, a system which can generate instrumental accompaniments to pair with input vocals! 📄arxiv.org/abs/2301.12662 🔊g.co/magenta/singso Work co-led by myself,

@antoine_caillon

, and

@ada_rob

as part of

@GoogleMagenta

and the broader MusicLM project 🧵

1:45

216.6K views

141

1,090

3,143

Looped Transformers as Programmable Computers Presents a framework for using transformer networks as universal computers by programming them with specific weights and placing them in a loop. arxiv.org/abs/2301.13196

199

A group of Spam cans empowered by AI to tell their tales. The story is created by a neural network tuned on a specially altered piggy version of Brave New World. In this novel each character is born into a caste in the same way industrial farm animals are born into their fate.

0:42

8.1K views

246

We're releasing our

hackathon 1st place project - "GPT is all you need for backend" with

@evanon0ping

@theappletucker

But let me first explain how it works:

102

615

2,306

One lesson from AI progress is it’s hard to predict the relative difficulties of various skills; one lesson from AI limitations is that humans are much more capable than we give ourselves credit for.

612

🔥 New (1h56m) video lecture: "Let's build GPT: from scratch, in code, spelled out." youtube.com/watch?v=kCc8Fm We build and train a Transformer following the "Attention Is All You Need" paper in the language modeling setting and end up with the core of nanoGPT.

543

3,630

21K

Banning ChatGPT for scientific writing is misguided. It's like banning calculators or spell checkers I find ChatGPT to be a very useful tool for generating a first version of a paragraph or refining an existing paragraph. I predict this policy will be "refined" in <1 yr.

296

Can't wait to share our new Text-to-Audio model, AudioLDM. 😆 This video shows the generation result with a simple text prompt: "A music made by xxx". More demos coming soon!😉 The paper will be available next Monday on arXiv! 😊 Our model will be open-sourced soon!😎

2:43

35.3K views

132

631

Use AI to Paint a Video alpha.genmo.ai 1️⃣ Upload a photo 📷

@genmoai

2️⃣ Paint over it and write a caption 3️⃣ One minute later, it's animated 🎥🖌️ New video inpainting model out now

0:23

9.8K views

163

ChatGPT (and others) generate very fluent (but not always truthful) text. Some worry that teachers, news-readers (like you!), and society in general will be swamped with AI-generated content. That's why we built DetectGPT, a method for detecting if text comes from an LM.

282

1,019

Unreal Engine 5 + 3D game assets generated with

@LumaLabs

new "Imagine" beta🤯 I used Text Prompts to generated all the 3D models on this coffee table scene... unlocking endless possibilities! #LumaAI #UnrealEngine5 #AI

@vibrantnebula

@LumaLabsAI

0:43

28.2K views

103

548

Use these 5 AI websites to make hours of effort disappear in minutes: (free) 🧵

258

1,717

4,993

a short teaser of what we’ve been up to lately ⁦

@stanford

⁩: here is an end-to-end policy running on low-cost arms!

0:18

54.4K views

101

790

Hey ! Long time no see in your TL 👋 Here's a prototype

@gradio

demo for a GPT Talking Portrait 🤗 It uses Whisper to let you ask GPT in your language, then One-Shot-Talking-Face will generate a talking portrait video from the answer ! Enjoy ! link: huggingface.co/spaces/fffilon

0:16

9.3K views

208

Multiple Animators from DreamWorks Animation have shared a behind-the-scenes look at the studio's Puss in Boots: The Last Wish, showing how the movie looked during production. See more: 80.lv/articles/a-beh #DreamWorks #animation #PussInBoots Video by

@Jeremy_Schaef

0:10

23.7K views

693

🚀 Fine-tune SantaCoder on code generation datasets with this repo: github.com/loubnabnl/sant A Google Colab by

@mrm8488

is also available. ✨ Bonus: we fine-tuned SantaCoder on Jupyter Notebooks to make it explain code huggingface.co/spaces/loubnab

3 AI websites you won't believe exist. SlidesAI.io Create presentation slides with AI UseBlackBox.io Convert plain English to code Krisp.ai Remove background voice, echo from all your calls

133

1,890

5,536