PerthMLGroup

[Peter McMahon @peterlmcmahon]

In our paper https://t.co/thETwnqw9u on the arXiv today, we present results from a study investigating the energy-efficiency advantage that could be achieved in executing state-of-the-art Transformer models on optical hardware.

[nixtla @nixtlainc]

🎉 We are thrilled to announce the release of the latest version of mlforecast a #Python library for Scalable #machinelearning 🤖 for #timeseries #forecasting

🚀 This version comes with exciting new features that are sure to make forecasting even more efficient and accurate

🧵

[John Nay @johnjnay]

Cyborgism Research Agenda

-Does not try to make GPT an agent
-Leverages GPT as a general simulator
-LLMs reason from scratch about any situation

-Advanced prompting interfaces
-Branches GPT chains of thought
-Injects variance into human thoughts

Post https://t.co/XUv0S4dcmx

[Paul Agapow @agapow]

This quietly slipped out yesterday. Briefly, the UK is traditinally a strong source of clinical trials but patient recruitment has fallen by 44% (!) due to slow & unpredictable set-up of research sites.

Independent review into UK clinical trials https://t.co/fQmrndiqtH

[Lewis Tunstall @_lewtun]

One exciting feature of our partnership with AWS is that we now have 1000+ GPUs to start training really large models at @huggingface 🔥🔥🔥!

We’ll be working hard to make closed models open, starting with LLMs and friends 🤓

Which closed models would you like to be open?

[John Nay @johnjnay]

Recitation-Augmented LLMs

-Usually: Retrieve relevant docs & give to LLM to answer

-Instead: Recite relevant passages from LLMs' own memory via sampling,
then produce final answer
-State-of-the-art on closed-book Q&A

Paper https://t.co/9UxJdIODaX
Code https://t.co/vXtc2HAmh4

[AI Breakfast @AiBreakfast]

German AI startup @Aleph__Alpha claims it will have a 300-billion parameter Large Language Model "Luminous-World" trained this year for “highly complex and critical applications”

(GPT-3 has 175 billion parameters)

[PyTorch @PyTorch]

We're excited to welcome FuseMedML to the PyTorch ecosystem!

FuseMedML is part of the BiomedSciAI organization which provides tools for AI-based accelerated discovery of biomarkers and molecules in the biomedical domain.

See the latest here: https://t.co/7d3ULIocVa

[Accepted papers at TMLR @TmlrPub]

Robust Hybrid Learning With Expert Augmentation

Antoine Wehenkel, Jens Behrmann, Hsiang Hsu, Guillermo Sapiro, Gilles Louppe, Joern-Henrik Jacobsen.

Action editor: Jasper Snoek.

https://t.co/MXigDAyAn4

#expert #models #modelling

[Kaggle @kaggle]

We've officially passed 200K public datasets!!! 👏👏

[Sebastian Raschka @rasbt]

"MarioGPT: Open-Ended Text2Level Generation through Large Language Models"
This looks like a pretty fun and creative project!
The best part, it's based on a distilled GPT-3 model and can be trained on a single GPU.

[Tansu Yegen @TansuYegen]

On a Welsh island, a seagull swallows a rabbit whole.

[Les Wright @LesLaboratory]

Supercontinuum in 100m of 50 micron Fiber

https://t.co/odFjnrSrQH

[Allen Downey @AllenDowney]

In the last week, three people on reddit/r/statistics have asked about testing whether a sample came from a Gaussian distribution.

The answer is that you should never test for normality. The result is a non-answer to the wrong question.

https://t.co/ZJIyvJHrKL

[Charly Wargnier @DataChaz]

.@TalarianHQ's `GPT for Sheets™` is the gift that keeps on giving! 🔥

Look at how easy it is to create personalized content with it, thanks to #GPT3's seamless integration! 🤯👇

Get the add-on here:
🔗https://t.co/3xUwkhBkc4

[Miguel Gaitan @miguelgaitan09]

@nonmayorpete I think this link will be useful: https://t.co/ouVsMMIcES
the paper: https://t.co/XNBFHoiKby
also they released a dataset:
https://t.co/uO43OuaGyS
it's a shame the actual library is not available but the research article is pretty good

[Sander Dieleman @sedielem]

Batch normalisation appears to be falling out of favour (probably for the best IMO, so many bugs end up being batchnorm bugs😬).

One area where it persists is GAN discriminators (e.g. in StyleGAN-T and VQGAN). Are there any other settings where batchnorm is still hard to avoid?

[Teklia @_Teklia_]

The first text line detection model for historical documents available on @huggingface : paper+code+models, all open-source https://t.co/SxG8hRlNd9 https://t.co/D8bEg4a3f7 #digitalhumanities @LitisLab

[fly51fly @fly51fly]

[CV] Learning Good Features to Transfer Across Tasks and Domains
P Z Ramirez, A Cardace, L D Luigi, A Tonioni, S Salti, L D Stefano [University of Bologna & Google] (2023)
https://t.co/fSwcZIFtCN
#MachineLearning #ML #AI #CV
[1/2]

[Pablo Villalobos 🔸 @pvllss]

Just published a literature review on scaling laws @EpochAIResearch! I collected a database of scaling laws for different tasks and architectures and reviewed dozens of papers in the scaling law literature.
Check out the database!
https://t.co/Hr3NTVQzwW
🧵