Another big step at this part of the story was moving from a more complicated proof based on functional derivatives to the simpler one based on convexity of KL divergences that we published. @dwf contributed this final step
RESEARCH
-
Non-Technical Description Of Generative Artificial Intelligence
By
–
A Non-Technical Description Of Generative Artificial Intelligence
#AI #AIio #BigData #ML #NLU #Futureofwork @oluskayacan @richardkimphd @MariaFariello @johnpearcenews5 @ResurgentAV @rohanmuralee @globaliqx http://
ow.ly/iejp30sInw5 -
Real-time AI forecasts represent next frontier advancement
By
–
it keeps getting better – real-time forecasts would be something!
-
Chicago builds quantum computing hub to rival Silicon Valley
By
–
Chicago is sick of seeing its groundbreaking research migrate to Silicon Valley when it’s time to become a business. For quantum computing, officials say, things will be different. https://
wsj.com/articles/chica
go-wants-to-build-the-silicon-valley-of-quantum-computing-8da25981?st=VTkjkP
… via @WSJ -
GraphCast: DeepMind’s Open-Source AI Weather Forecasting Model
By
–
Check out the codebase here: https://
github.com/google-deepmin
d/graphcast
… -
DeepMind Open Sources GenCast Weather Forecasting Model
By
–
DeepMind is on fire today, fully open sourced GenCast – SOTA diffusion based weather forecasting (ensemble) model! can predict temperature 15 days ahead in just 8 minutes powered by TPU v5e Code and weights open source! (although in a google cloud bucket)
-
HellaSwag Evaluation Script for Language Models Released
By
–
You can also find a HellaSwag evaluation script that works with one large batch item, EOT separators, and document_mask here: https://
github.com/alexjc/nanogpt
-speedrun/blob/master/eval_gpt2.py
… It's setup to use TokenMonster too, so you can check the scores on downstream benchmarks easily hopefully. -
NanoGPT Achieves 40% Training Efficiency Boost Via Tokenizer
By
–
New Submission On NanoGPT Speed-Run! Improved training token efficiency by ~40% to reach target HellaSwag score by switching the vocabulary and tokenizer only. It now takes 1050 steps instead of 1750 steps; current record on 11/24/24 labelled in the graph as "TikToken
-
Hazard and Survival in AI Systems
By
–
Hazard and Survival https://
bit.ly/4fXRrR3
#AI #MachineLearning #DeepLearning #LLMs #DataScience