RESEARCH - AI Dynamics

Beyond Metrics: Evaluating Research Quality in AI

By

–

06 January 2023 15h37

Worth repeating. If the only positive things you can say about a researcher is that they have many papers or citations, then either:
a) you don't know their work very well
b) they haven't done much interesting work
I always push for justification of quality beyond metrics.

→ View original post on X — @thegautamkamath

6 January 2023

Perspectives in Artificial Intelligence: Vision of Experts

By

@actuiafr

–

06 January 2023 14h24

Perspectives for artificial intelligence: discover the vision of renowned experts in the latest issue of #ActuIA https://actuia.com/actualite/perspectives-pour-lintelligence-artificielle-decouvrez-la-vision-dexperts-de-renom-dans-le-dernier-numero-dactuia/
… #AI #artificialintelligence #health #energy #defense #spain

→ View original post on X — @actuiafr

6 January 2023

YOLOv5 Object Detection Models: Performance and Optimized Inference Speed

By

@learnopencv

–

06 January 2023 14h11

YOLOv5 object detection models are well known for their excellent performance and optimized inference speed.🚀
▶️ In case you missed it, check out the full read on YOLOv5 Instance Segmentation – https://t.co/XPIV4ntDjZ #yolov5 #segmentation #computervision #deeplearning #ai pic.twitter.com/yQfGitP3D5
— Satya Mallick (@LearnOpenCV) 6 janvier 2023

YOLOv5 object detection models are well known for their excellent performance and optimized inference speed. In case you missed it, check out the full read on YOLOv5 Instance Segmentation – https://
learnopencv.com/yolov5-instanc
e-segmentation/
… #yolov5 #segmentation #computervision #deeplearning #ai

→ View original post on X — @learnopencv

6 January 2023

OpenAI Beta Duration and Data Collection Strategy

By

@thatroblennon

–

06 January 2023 13h55

Really hard to guess. Some of the original OpenAI betas lasted months or maybe even longer than a year if I recall. But with the technology being where it's at, they can start charging anytime they want. Right now they want more training data for how it gets used instead.

→ View original post on X — @thatroblennon

6 January 2023

GPT-2 Output Detector Limitations and AI Text Detection Evolution

By

@dotcsv

–

06 January 2023 10h13

No realmente. GPT-2 Output Detector, como bien indica su nombre, funciona para GPT-2. Fue una tecnología que OpenAI sacó cuando publicaron GPT-2 hace años, y aunque funcionaba, manipulando un poco el texto se podía engañar al detector. Ya con GPT-3 y demás IAs ni lo intentaron.

→ View original post on X — @dotcsv

6 January 2023

Biggest Tech Innovations and Breakthroughs of 2022

By

@bernardmarr

–

06 January 2023 9h03

The Biggest Tech Innovations And Breakthroughs Of 2022 In this article, we take stock and look at the biggest #technology #innovations or technological #breakthroughs of #2022. https://
linkedin.com/pulse/biggest-
tech-innovations-breakthroughs-2022-bernard-marr
…

→ View original post on X — @bernardmarr

6 January 2023

Fine-tuning is key to AI model optimization

By

@levelsio

–

06 January 2023 7h36

But yes the key is finetuning A LOT

→ View original post on X — @levelsio

6 January 2023

PPO Explanation Gaps and InstructGPT Fine-tuning Data Requirements

By

@swyx

–

06 January 2023 5h38

thanks man its a good thread. i feel like people dont explain PPO very well in these threads, thats the main part that always feels handwavy. i’d also like an idea of order of magnitude of finetune data needed for instructgpt but havent seen good numbers

→ View original post on X — @swyx

6 January 2023