polybox

Automate it.

[Jim Fan @DrJimFan]

GPT3 is powerful but blind. The future of Foundation Models will be embodied agents that proactively take actions, endlessly explore the world, and continuously self-improve. What does it take? In our NeurIPS Outstanding Paper “MineDojo”, we provide a blueprint for this future:🧵

[Michael Black @Michael_J_Black]

With LLMs for science out there (#Galactica) we need new ethics rules for scientific publication. Existing rules regarding plagiarism, fraud, and authorship need to be rethought for LLMs to safeguard public trust in science. Long thread about trust, peer review, & LLMs. (1/23)

[Papers with Code @paperswithcode]

🪐 Introducing Galactica. A large language model for science.

Can summarize academic literature, solve math problems, generate Wiki articles, write scientific code, annotate molecules and proteins, and more.

Explore and get weights: https://t.co/jKEP8S7Yfl

[Marcel Schwittlick @schwittlick_]

Computer and Aesthetics
Searching for the Art of Artificial Intelligence
Hiroshi Kawano, 1984

[Misha Laskin @MishaLaskin]

In our new work - Algorithm Distillation - we show that transformers can improve themselves autonomously through trial and error without ever updating their weights.

No prompting, no finetuning. A single transformer collects its own data and maximizes rewards on new tasks.

1/N

[Michael Black @Michael_J_Black]

Today at #ECCV2022 I'm giving a talk on "The risks and opportunities of academic startups" in the Industry Track. Hall KLM (1st floor) 15:45. I hope to make it interactive so come with questions.

[Classical Studies Memes @CSMFHT]

Well that's me sorted for Halloween

[Frank Hutter @FrankRHutter]

Please see our blog post https://t.co/WtEZ06Q8JF for details & code. Also, we just got an oral at the #NeurIPS table representation learning workshop https://t.co/XOI1n3qyst 🎉Joint work with my outstanding students @noahholl, @SamuelMullr and @KEggensperger. 6/6

[Frank Nielsen @FrnkNlsn]

A selection of books on Algorithmic Differentiation (AD) also called Automatic Differentiation:
The backbone toolbox for differentiable programming and deep learning
https://t.co/uUGA0UqdPr

[Mert R. Sabuncu 🤖🩻⚕️ @mertrory]

Some advice I give my graduate students about how to keep on top of the literature in their field. A 🧵: 1/

[Abi Aryan @GoAbiAryan]

This has been such an excellent year for software system design in ML. So, I compiled a list of some of my favorite papers 📜in MLOps.

Here are some of my favorite ones till date⤵️

[Gale Na @poisonjr]

"overnight oats" "gig economy" "tiny home" you are a peasant who eats gruel

[AI Pub @ai__pub]

Shreya Shankar releases an excellent real-world MLOps interview study with 18 ML engineers:

"When considered together, these responsibilities seem staggering -- how does anyone do MLOps, what are the unaddressed challenges, and what are the implications for tool builders?"

9/15

[AI Pub @ai__pub]

Since OpenAI released Whisper a three days ago, there have been some outstanding paper + code walkthroughs on AI Twitter + AI YouTube.

A few below if you haven't seen them:

1/4

[Jonathan Whitaker @johnowhitaker]

New notebook: "Grokking Stable Diffusion"
Hopefully useful for anyone looking for content that goes deeper than just how to run the model!
https://t.co/O7GALcgTxX
Covering:
- The diffusion loop
- Messing with text embeddings
- img2img
- arbitrary guidance
More deets in thread 🧵

[Javier Blas @JavierBlas]

Fresh record high prices for 🇩🇪 and 🇫🇷 electricity:

German 1-year forward: €725 per MWh
French 1-year forward: €870 per MWh

The 2010-2020 average was around €41 per MWh

Russian energy weapon; French nuclear crisis; low wind production. Drought-hit hydro. #EnergyCrisis

[james tom @jestom]

one thing I like about wfh is I can have my gastrointestinal issues at my own pace

[Animesh Garg @animesh_garg]

We have now released code for SHAC from #ICLR2022

Differentiable Simulation (dFLEX) for Model-Based RL for sample efficient policy learning.

https://t.co/hpaEl7n5eW

Thanks to the efforts of many folks, especially @viktor_m81 and @xujie7979

[Randall Munroe @xkcd]

THREAD: The evolution of Pokémon cards through history, as generated by DALL·E 2

For starters, here’s what DALL·E 2 thinks 21st century Pokémon cards look like, using prompts like “A Pokémon card from 2001”