Bernhard Liebl retweeted
Today we are launching pixi! It's our new package manager for the conda ecosystem by @prefix_dev. It comes with powerful lockfiles out of the box, cross-platform workflows and the speed of Rust.
Bernhard Liebl retweeted
A 14-line Python script using gzip outperforming a 345m parameter transformer model is probably the most hilarious result I've seen all year.
aclanthology.org/2023.findin…
Bernhard Liebl retweeted
This is a bigger deal than it sounds. IMO as a field we focus too much on hypothetical long-term challenges and not enough on current challenges that we can find solutions for safer & more ethical AI. Are you using safetensors by @narsilou yet?
huggingface.co/blog/safetens…
Bernhard Liebl retweeted
After over a year of development, I am excited to release the first stable version of *nanobind*, a tiny library for efficient C++/Python bindings. Much has happened since the first announcement: CPU/GPU data exchange with tensor frameworks, Eigen dense/sparse matrices, .. 1/3
Bernhard Liebl retweeted
a big deal: @elonmusk, Y. Bengio, S. Russell, @tegmark, V. Kraknova, P. Maes, @Grady_Booch, @AndrewYang, @tristanharris & over 1,000 others, including me, have called for a temporary pause on training systems exceeding GPT-4 futureoflife.org/open-letter…
Bernhard Liebl retweeted
2.5 months later...in collaboration with @openAI , ChatGPT gets its "Wolfram superpowers"!
writings.stephenwolfram.com/…
Bernhard Liebl retweeted
We have studied #GPT4 for several months. Initial curiosity has been deepened by intriguing findings that we share here arxiv.org/abs/2303.12712 Exciting time to pursue deeper understandings of emergent “sparks” of powers of generalization, abstraction, synthesis. @MSFTResearch
At @MSFTResearch we had early access to the marvelous #GPT4 from @OpenAI for our work on @bing. We took this opportunity to document our experience. We're so excited to share our findings. In short: time to face it, the sparks of #AGI have been ignited.
arxiv.org/abs/2303.12712
Bernhard Liebl retweeted
Scalable Viewing oder doch eher Viewing at Scale? Nach dem Vortrag gestern von @BenKrautter könnte man die Begrifflichkeit evtl. nochmal überdenken? 🤔 (aus dem spannenden Vortrag von @8urghardt auf der #DHd2023 gesendet)
Bernhard Liebl retweeted
A first look at the UI of our Workflow Editor which will allow users to craft simple and complex OCR workflow within OCR4all as well as in the stand-alone application NodeFlow.
Bernhard Liebl retweeted
Async programming is so cool. With just a few minutes of work you can have a program that hangs forever and is impossible to debug
Bernhard Liebl retweeted
next level storage reddit.com/r/programmerhumor…
Bernhard Liebl retweeted
If you installed PyTorch-nightly on Linux between Dec. 25 and Dec. 30, uninstall it and torchtriton immediately and use the latest nightly binaries.
Read the security advisory here: pytorch.org/blog/compromised…
Bernhard Liebl retweeted
It's good that PyTorch found this so quickly.
However it makes me worry that many packages share this issue without realizing. You could easily clone pip wheels and add exploits without being obvious.
The fragmentation of the ecosystem (see xkcd) is seeing some consequences...
If you installed PyTorch-nightly on Linux between Dec. 25 and Dec. 30, uninstall it and torchtriton immediately and use the latest nightly binaries.
Read the security advisory here: pytorch.org/blog/compromised…
Bernhard Liebl retweeted
I am THRILLED to announce that the #ConceptArtAssociation has released a #gofundmecampaign to fund multiple actions, so we artists can stand up for our rights and be heard in the public and in DC! Art friends please donate, SHARE this link. Let's do this!
gofund.me/2df3dc07
Bernhard Liebl retweeted
MetaPortrait: Identity-Preserving Talking Head Generation with Fast Personalized Adaptation
abs: arxiv.org/abs/2212.08062
project page: meta-portrait.github.io/
Bernhard Liebl retweeted
1/Large language models like Galactica and ChatGPT can spout nonsense in a confident, authoritative tone. This overconfidence - which reflects the data they’re trained on - makes them more likely to mislead.
Bernhard Liebl retweeted
Dear @TwitterSupport, as much as we enjoy our excellent bilateral relations and close alphabetical vicinity with Nigeria, we would much appreciate if you could label us as Norway😉
P.S. That also goes for Prime Minister @jonasgahrstore and Foreign Minister @AHuitfeldt 🇳🇴
Bernhard Liebl retweeted
Terrific work by @AinslieJstone from @TheEconomist! A perfect illustration of Simpson's paradox:
* Overall in Europe, WARMER countries have HIGHER winter excess mortality. I suspect, due to the lack of heating.
* But WITHIN each country, colder winter => more excess deaths. 🥶
Replying to @AinslieJstone
It’s not just sub-zero temperatures either. In fact, warmer countries actually tend to have more additional deaths over winter, and are more affected by small deviations in winter temperature (more on this in last week’s Off the Charts newsletter: economist.com/newsletters/of…)








































