News

How a big shift in training LLMs led to a capability explosion
When someone starts a new job, early training may involve shadowing a more experienced worker and observing what they do ...
Novel model optimization algorithm improves robustness of near-infrared spectroscopy models
A research team from the Hefei Institutes of Physical Science of the Chinese Academy of Sciences has proposed a novel model optimization algorithm—External Calibration-Assisted Screening (ECA)— that ...
IEEE23d
COMQ: A Backpropagation-Free Algorithm for Post-Training Quantization - IEEE Xplore
Post-training quantization (PTQ) has emerged as a practical approach to compress large neural networks, making them highly efficient for deployment. However, effectively reducing these models to their ...
IEEE2mon
A New Correntropy-Based Conjugate Gradient Backpropagation Algorithm for Improving Training in Neural Networks - IEEE Xplore
Published in: IEEE Transactions on Neural Networks and Learning Systems ( Volume: 29 , Issue: 12 , December 2018 ) ...
EurekAlert!1y
Training algorithm breaks barriers to deep ph | EurekAlert!
Reference: Ali Momeni et al, Backpropagation-free training of deep physical neural networks. Science 0, eadi8474 DOI: 10.1126/science.adi8474 ...
InfoQ2y
Deep Learning Pioneer Geoffrey Hinton Publishes New Deep Learning Algorithm
The Forward-Forward algorithm (FF) is comparable in speed to backpropagation but has the advantage that it can be used when the precise details of the forward computation are unknown.