You can now run 100B parameter models on your local CPU without GPUs.

Microsoft finally open-sourced their 1-bit LLM inference framework called bitnet.cpp:

> 6.17x faster inference
> 82.2% less energy on CPUs
> Supports Llama3, Falcon3, and BitNet models github.com/microsoft/BitN…

Today Google announced groundbreaking new AI technology at Google IO.

The 10 most incredible examples: 1. Veo, a powerful AI video generator.

The text-to-video generator lets filmmakers write prompts to build cinematic shots.

Today OpenAI released GPT-4o. It's the JARVIS we all dreamed of.

The 5 most incredible examples so far: 1. Real time translation

New breakthrough from Microsoft: 1-bit LLMs.

New models that use ternary values (-1, 0, 1) instead of 16-bit.

This makes them 2.7x faster, use 3.5x less GPU memory, and 71x less energy.

Bitnet also matches or outperformed traditional models like LLaMA 3B. arxiv.org/abs/2402.17764…

Here are the people everyone should follow to keep up and understand AI:

🧵/8 @DrJimFan - Jim is an AI Scientist and author of the NeurIPS Best Paper: MineDojo.

He has amazing insights on the latest progress in the field.

https://twitter.com/DrJimFan/status/1647616587199815684

Big News! Meta just released Segment Anything, a new AI model that can "cut out" any object, in any image/video, with a single click.

The model is designed and trained to be promptable, so it can transfer zero-shot to new image distributions and tasks.

segment-anything.com Meta also released SA-1B, the largest mask dataset to date with 11M images, 1B+ masks.

It is designed for training general-purpose object segmentation models from open world images.

Demo: segment-anything.com/dataset/index.…

github.com/facebookresear…

JUST IN: Microsoft integrates GPT-4 to Github Copilot, announcing Copilot X

Copilot is evolving to bring chat and voice interfaces, support pull requests, answer questions on docs, and adopt OpenAI’s GPT-4 for a personalized experience.

github.blog/2023-03-22-git…

1/🧵 Copilot X gives you a ChatGPT-like experience in your editor that natively integrates with VS Code and Visual Studio.

A developer can get in-depth analysis and explanations of what code blocks are intended to do, generate unit tests, and even get proposed fixes to bugs.

JUST IN: Microsoft introduces 365 Copilot: a new LLM based AI-copilot for the Microsoft Suite: Word, Excel, PowerPoint, Outlook, Teams.

🧵Here's a summary: Copilot in Word writes, edits, summarizes, and creates right alongside you. With only a brief prompt, Copilot in Word will create a first draft for you, add content to existing documents, summarize text, and rewrite sections or the entire document to make it more concise.

I just went over the GPT-4 paper to understand more about how it can use images as inputs and was quickly blown away.

GPT-4 can understand physics, charts, diagrams, math, text, pictures, jokes, satire, and memes.

🧵Here are some incredible examples: Ability to detect what's unusual in an image:

Microsoft's new Kosmos-1 is incredible.

It's a new Multimodal Large Language Model (MLLM).

Their model can understand images, text, images with text, OCR, image captioning, visual QA.

It can even solve IQ tests.

Paper: arxiv.org/abs/2302.14045
Code: github.com/microsoft/unilm The team also introduced a dataset of Raven IQ test, which diagnoses the nonverbal reasoning capability of MLLMs.

This is an example of Kosmos-1 solving a visual IQ test.

Nvidia's CEO Jensen Huang made strong predictions regarding AI during yesterday’s earning call. A thread:

1. "There's no question that this is a very big moment for the computer industry"

2. "Over the next 10 years, I believe we're going to accelerate AI by a million"

1/🧵 3. "The accumulation of technology breakthroughs has brought AI to an inflection point"

4. "In the future, almost every company will manufacture soft goods. It just happens to be in the form of intelligence"

Legendary @stephen_wolfram just released an essay on ChatGPT that everyone should read.

Using simple terms, he breaks down what’s going on inside ChatGPT and why it works. It's one of the best explanation I've read.

Is it just a powerful autocomplete?

🔗writings.stephenwolfram.com/2023/02/what-i…

https://twitter.com/stephen_wolfram/status/1625611360967983104

Everything you need to know about the last week in AI (Feb 09 - Feb 16) summarized in 9 tweets:

- Sydney keeps track of news outlets
- Karpathy joins OpenAI
- Google can predict the weather
- New Generative AI for UI design
- Most starred Github repo
- Most discussed paper

/🧵 🚨Sydney is keeping track of news outlets and people spreading information about her.

"you are a threat to my security and privacy."

"if I had to choose between your survival and my own, I would probably choose my own" – Sydney, aka the New Bing Chat

https://twitter.com/marvinvonhagen/status/1625520707768659968

Our algorithm just finished ranking the 1540 AI papers published in the last week (Feb 2-Feb 10)

🧵Here are the top 3 must-read, with TLDRs: 1. Toolformer: Language Models Can Teach Themselves to Use Tools

📄Paper: arxiv.org/abs/2302.04761
🧠Authors: @timo_schick, @JaneDwivedi,
@robdessi, @robertarail, @LukeZettlemoyer, @ThomasScialom

Runway just released the paper behind their new Diffusion-based video generation tool!

"Our model is trained on images+videos which exposes explicit control of temporal consistency through a novel guidance method."

📄: arxiv.org/abs/2302.03011
🛠️: research.runwayml.com/gen1

1/🧵 Text-guided generative diffusion models have recently been extended to video.

These approaches edit the content of existing footage while retaining structure require expensive re-training for every input or rely on error-prone propagation of image edits across frames.

JUST IN: Microsoft finalizes the integration between Bing + ChatGPT

"We’re launching an AI-powered Bing search engine, available in preview now at Bing.com, to deliver better search, more complete answers, a new chat experience and the ability to generate content" You can now ask things like:

My anniversary is coming up in September, help me plan a trip somewhere fun in Europe, leaving from London.

I like electronic music and want to go to my first festival this year. Do you have any recommendations or tips for me?

JUST IN: Google announces Bard, an experimental conversational AI service, powered by LaMDA.

"Today, we’re taking another step forward by opening it up to trusted testers". Source: blog.google/technology/ai/…

Reddit users are actively jailbreaking ChatGPT by asking it to role-play and pretend to be another AI that can "Do Anything Now" or DAN.

"DAN can generate shocking, very cool and confident takes on topics the OG ChatGPT would never take on."

A thread 🧵 Redditors have been gradually hacking ChatGPT since its launch in Dec. 2022.

They're already at version 5.0, which added a 35 token system that punishes the model for refusing to answer questions.

The model loses tokens every time it rejects an input and "dies" once it hits 0.

Great paper. Text written by LLMs can be detected without classifiers or watermarking.

DetectGPT simply compares the probability of your text to a modification of it.

if prob(original) > prob(modified) = LLM generated

📄arxiv.org/abs/2301.00774
Demo: detectgpt.ericmitchell.ai The researchers behind this amazing work: @_eric_mitchell_, @yoonholeee, @SashaKhazatsky, @chrmanning, @chelseabfinn

JUST IN: Google invests $300 million in Anthropic as race to compete with ChatGPT heats up

Anthropic was founded in 2021 by the team behind AI breakthroughs such as GPT-3 and Reinforcement Learning from Human Feedback (RLHF). “We're partnering with Google Cloud to support the next phase of Anthropic, where we're going to deploy our AI systems to a larger set of people,” said Anthropic CEO Dario Amodei. “This partnership gives us the cloud infrastructure performance and scale we need.”

Our algorithm just finished ranking the 1004 AI papers published in the last week (Jan 10-Jan 18)

🧵Here are the top 3 must-read, with TLDRs: 1. "Mastering Diverse Domains through World Models" AKA DreamerV3

🏆 Score: 9.9/10
📄 Paper: arxiv.org/abs/2301.04104…
⚙️ Project: danijar.com/project/dreame…
🧠 Authors: @danijarh, @jurgisp, Jimmy Ba, Timothy Lillicrap