Latest posts tagged with #optimizers on Bluesky
New #J2C Certification:
AdaCubic: An Adaptive Cubic Regularization Optimizer for Deep Learning
Ioannis Tsingalis, Constantine Kotropoulos, Corentin Briat
https://openreview.net/forum?id=pZBQ7J37lk
#adacubic #regularization #optimizers
How iteration composition influences convergence and stability in deep learning
Benoit Dherin, Benny Avelin, Anders Karlsson, Hanna Mazzawi, Javier Gonzalvo, Michael Munn
Action editor: Konstantin Mishchenko
https://openreview.net/forum?id=GZCBM2Yo3a
#iteration #batches #optimizers
GLOV: Guided Large Language Models as Implicit Optimizers for Vision Language Models
Muhammad Jehanzeb Mirza, Mengjie Zhao, Zhuoyuan Mao et al.
Action editor: Masashi Sugiyama
https://openreview.net/forum?id=kZLANTp6Vw
#vision #optimizers #recognition
New #J2C Certification:
Low-rank Momentum Factorization for Memory Efficient Training
Pouria Mahdavinia, Mehrdad Mahdavi
https://openreview.net/forum?id=W3D3TVo9a3
#optimizers #sgd #benchmarks
New #J2C Certification:
Celo: Training Versatile Learned Optimizers on a Compute Diet
Abhinav Moudgil, Boris Knyazev, Guillaume Lajoie, Eugene Belilovsky
https://openreview.net/forum?id=SLqJbt4emY
#optimizers #optimizer #hyperparameter
Low-rank Momentum Factorization for Memory Efficient Training
Pouria Mahdavinia, Mehrdad Mahdavi
Action editor: John Timothy Halloran
https://openreview.net/forum?id=W3D3TVo9a3
#optimizers #sgd #benchmarks
Celo: Training Versatile Learned Optimizers on a Compute Diet
Abhinav Moudgil, Boris Knyazev, Guillaume Lajoie, Eugene Belilovsky
Action editor: Vikas Sindhwani
https://openreview.net/forum?id=SLqJbt4emY
#optimizers #optimizer #hyperparameter
Optimizer Noise Shapes Model Merging Success in Neural Networks
Effective noise scale—combining learning rate, weight decay, batch size and augmentation—predicts model‑merging success, with a non‑monotonic optimum. Read more: getnews.me/optimizer-noise-shapes-m... #modelmerging #effectivenoisescale #optimizers
Simpler Optimizers Boost Robustness in Large Language Model Unlearning
Gradient‑free optimizers improve LLM unlearning robustness; a hybrid using both first‑ and zeroth‑order steps performed best on MUSE and WMDP benchmarks. Read more: getnews.me/simpler-optimizers-boost... #llmunlearning #optimizers
Per-Example Gradient Statistics Open New Paths for Optimizer Design
Research shows per‑example gradient stats have negligible overhead vs mini‑batch gradients, and applying the sign in SignSGD after aggregation preserves signal‑to‑noise ratio. Read more: getnews.me/per-example-gradient-sta... #optimizers #signsgd
AuON optimizer offers linear‑time semi‑orthogonal update alternative
AuON optimizer delivers linear‑time (O(n)) momentum updates, matching the performance of AdamW and Muon on vision and language benchmarks. The code is available on GitHub. Read more: getnews.me/auon-optimizer-offers-li... #auon #optimizers
Conda optimizer accelerates large language model training
The new Column-Normalized Adam (Conda) optimizer, released on 29 Sep 2025, speeds LLaMA training up to 2.5× faster than AdamW, with code available on GitHub. Read more: getnews.me/conda-optimizer-accelera... #conda #llm #optimizers
Gradient-Based Optimizers Reduce Cycle Skipping in Waveform Inversion
Researchers applied gradient‑based optimizers with larger learning rates to FWI, showing they can escape cycle‑skipping in benchmarks and real field data, despite missing low‑frequency content. getnews.me/gradient-based-optimizer... #fwi #optimizers
#Economists and others are used to building #forecasts on the assumption that the agents involved in what they’re forecasting are #rational #optimizers. Makes it difficult when the most important actor is a #narcissist with an inexhaustible need for ego gratification.
Implicit Bias and Fast Convergence Rates for Self-attention
Bhavya Vasudeva, Puneesh Deora, Christos Thrampoulidis
Action editor: Han Bao
https://openreview.net/forum?id=pKilnjQsb0
#optimizers #attention #adaptive
Relationship between Batch Size and Number of Steps Needed for Nonconvex Optimization of Stochast...
Yuki Tsukada, Hideaki Iiduka
Action editor: Alec Koppel
https://openreview.net/forum?id=pqZ6nOm3WF
#optimizers #sgd #optimization
Habrá que apuntar este nuevo nicho, parece que está revolucionando el sector de las baterías en Reino Unido, Alemania, California...
Llegan los #optimizers a España, un nuevo negocio que consigue que las #baterías conectadas a la red eléctrica sean más rentables
www.elespanol.com/invertia/emp...
This is something I haven't seen before. I'm interested to hear what industry people think about this. #solar #optimizers
Stochastic Re-weighted Gradient Descent via Distributionally Robust Optimization
Ramnath Kumar, Kushal Alpesh Majmundar, Dheeraj Mysore Nagaraj, Arun Suggala
Action editor: Mathurin Massias
https://openreview.net/forum?id=KCf5CLAXZq
#sgd #imagenet #optimizers
An amazing #wcgo2019 conference banquet tonight filled with delicious food and fabulous conversations! Many thsnkbto the organizers. Not every day one gets to meet #optimizers from countries such as #Kazakhstan, #Morrocco, #Benin, #Algeria, and #Iran, among others.
Uh oh, @barts is showing his brain again! #eweek #optimizers
En @Atopos42 heeft weer betere dingen te doen dan te luisteren naar @tonw... #eweek #optimizers #zoleerjehetnooit
Denk je alles wel gehad te hebben, kom je @Atopos42 tegen! #eweek #optimizers
