AndyEdmondson @AndyEdmondson@qoto.org

Training update: Atlas learned to stand still! #reinforcementlearning #machinelearning #MLAgents #artificialintelligence

Robotic Skill Acquisition via Instruction Augmentation with Vision-Language Models
abs: https://arxiv.org/abs/2211.11736
project page: https://instructionaugmentation.github.io/

Holy moly this thingy is sick!

Good morning 🙂

And here’s yesterday’s Stable Diffusion prompt in the Pratchett title series: “equal rites” (see pinned post for standard modifiers added).

The one with the spaceship is from the RetroSF model, but the rest are from standard Stable Diffusion.

Not sure why everything has a battle/fighting vibe for “equal rites” but maybe the AI thinks that it’s “equal rights” and that that’s a struggle?

#AIArtwork #StableDiffusion #DeepLearning #AI #DiscWorld

Today’s, or rather yesterday’s, Stable Diffusion prompt in the Terry Pratchett title series was simply — “mort” (with my standard modifiers — see pinned post).

I love short prompts because you never know what you’ll get or how the AI will interpret the prompt.

In this case, most of the ones I got were horror-y or had some sort of a death’s head motif. But there were some that were different and those were the ones I liked the best 🙂

#AIArt #StableDiffusion #DeepLearning #CV #AI #DiscWorld

Facebook (sorry: Meta) AI: Check out our "AI" that lets you access all of humanity's knowledge.

Also Facebook AI: Be careful though, it just makes shit up.

This isn't even "they were so busy asking if they could" --- but rather they failed to spend 5 minutes asking if they could.

#AL #ML #MathyMath #Bullshit #NLP #NLProc #AIhype

I am dead, I can’t believe this AI intro text I just got. If you thumb through it, there are all the usual suspects like breadth-first search and probability.

Then you open the first chapter and it goes HARD on the current state of things.

Then it’s back to mathematical notation like nothing happened.

Deep Whole-Body Control: Learning a Unified Policy for Manipulation and Locomotion

📜 https://arxiv.org/abs/2210.10044

❓ Task: Joint control of a quadruped's legs and attached arm.

💡 Idea: Regularization allowing joint training of privileged and sensor-based encoders in simulation. For rewards, initially attribute manipulation to arm and locomotion to legs, and anneal over training.

📈 Result: Successful interaction and coordination between legs and arms on real robot.

#Robotics #ReinforcementLearning

If you are an academic working in the UK, consider joining the University and College Union (UCU, @ucu@twitter.com, https://ucu.org.uk/join) -- the process is extremely quick and simple! #union #UCU #UK #academia

”Global Witness found more than 600 people at the talks in Egypt are linked to fossil fuels.
That's more than the combined delegations from the 10 most climate-impacted countries.”

"If you want to address malaria, you don't invite the mosquitoes”
#COP27

https://www.bbc.com/news/science-environment-63571610

my Likert scale:

⭐️⭐️⭐️⭐️⭐️: lovert
⭐️⭐️⭐️⭐️: likert
⭐️⭐️⭐️: whatevert
⭐️⭐️: dislikert
⭐️: hatert

Most #AI, #ML research assumes the first configuration in the figure below. We take it for granted that a human will fully delegate a task to the #AI #system

That couldn't be further from reality. For a long time, humans and #AI/#ML systems will have to work together in various configurations.

#CollaborativeHumanAISystems looks at what is typically overlooked in #AI, #ML - how do we bring humans in the #AI loop.

A more wavy cyber Mastodon. #AiArt #StableDiffusion

My work was recently recognized with #MichiganAI, #UMichCSE as #AI #RisingStar.

I put forth an agenda of #CollaborativeHumanAISystems - #IntelligentSystems that are designed to model, reason, and learn about their human partners -supporting them in their goals and pursuits.

It is #AI, #ML research but extremely #HumanCentered. We start with studying what is the human trying to do and what intelligent support do they need.

A thread below -

7 Data-based and Artificial Intelligence Projects to Help Us Fight Climate Change.
#AI #ClimateChange #Climate #science
https://earth.org/artificial-intelligence-projects/

Holy hell the twitter exodus is **still** accelerating. The influx of new users is up about 33% from yesterday, our biggest day so far. We are now seeing over 1,300 new users every day on QOTO!

Since this has grown a lot since our first #introduction , FYI:

The Gradient is a non-profit and volunteer-run organization with the mission of facilitating discussion within the AI community. We were founded in 2017 by a group of students and researchers at the Stanford AI Lab. Our current projects include The Gradient Magazine, The Gradient Podcast, The Update newsletter, and Sigmoid Social (the site you are currently on).

You can read more about us here: https://thegradient.pub/about/

Hi, this is Greta. You are welcome to follow me here, if you prefer planet earth to planet mars.

So that’s how @carbon_re comes into the picture.

We develop #AI based tools that enable these plants to improve their operations, aiming to get to optimal control policies. Cement produced in an optimally controlled plant generates significantly less CO2 than an average human-controlled plant, and there are only ~3000 of them to fix up!

It is more #sustainable, safer for workers, and pushes the industry to improve across the board. Improvements are good, and we need more of them, faster!

5/🧵