Running those old configs today would be hard. They need pylearn2, pylearn2 needs an old version of theano, old versions of theano need old versions of CUDA, old versions of CUDA need old GPUs.
MACHINE LEARNING
-
First GAN Config File: Historical Artifact and Reproducibility Challenge
By
–
After I graduated I made my personal private repo from grad school public. So I think the first ever GAN config file was this one: https://
github.com/goodfeli/galat
ea/blob/master/adversarial/adversary.yaml
… Full provenance is not preserved (what pylearn2 version, theano version, CUDA version, libc version etc was I running?) -
ML NLP LLMs RAG Newsletter Daily Tips Tutorials
By
–
If you’re interested in:
– ML/NLP
– LLMs
– RAG My Newsletter → http://
unwindai.substack.com Everyday, I share tip & tutorials on above topics on X and my newsletter. -
Quantization in Depth: Compressing ML Models Efficiently
By
–
Have you used quantization with an open source machine learning library, and wondered how quantization works? How can you preserve model accuracy as you compress from 32 bits to 16, 8, or even 2 bits? In our new short course, Quantization in Depth, taught by @huggingface's… pic.twitter.com/7sj1Ivv5a4
— Andrew Ng (@AndrewYNg) 6 mai 2024Have you used quantization with an open source machine learning library, and wondered how quantization works? How can you preserve model accuracy as you compress from 32 bits to 16, 8, or even 2 bits? In our new short course, Quantization in Depth, taught by @huggingface
's -
Data Soup: How AI Processes Training Information
By
–
Souped up in the sense of turned into data soup.
-
Scalable Self-Play with LLMs: Current Research and New Approaches
By
–
What's the closest researchers have gotten to scalable self-play w/ LLMs? Had an idea I think is unique and easily implementable, curious if there is anything I should take a look at.
-
AI accuracy depends on training data quality and enterprise development
By
–
Dataconomy is spot on. The accuracy of AI depends on the data it's trained on. Incorporating enterprise data in AI/ML models is difficult and time-consuming. That's why AI data development is so critical. Read the full story: https://
buff.ly/4a3VqrA -
LeRobot: Hugging Face launches first robotics library
By
–
Welcome LeRobot! the first robotics library at Hugging Face Over the past months, we saw impressive research breakthroughs in robotics (ALOHA, diffusion policies, UMI and so many) enabling robots behaviors previously thought impossible to train with limited quantity of data
-
DistBelief Legacy: Model and Data Parallelism in Modern Neural Networks
By
–
I'm still a bit surprised that you think the system (DistBelief) that first combined large scale model- and data- parallelism with backprop-based training of very large neural nets is a "dead end", given that nearly every large-scale model that is trained today combines these