darts
PulseAugur coverage of darts — every cluster mentioning darts across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
New LESS method slashes neural architecture search time to minutes
Researchers have developed LESS (Lightweight Evolutionary Supernet Search), a novel method for neural architecture search that significantly reduces computation time. By combining a brief warm-up with a discrete search …
-
New DARTS technique improves decoder LLM merging with entropy-weighted loss
Researchers have developed a new technique called DARTS (Decoder-Aware Representation Tuning via Surgery) to improve model merging for decoder-based large language models. Unlike previous methods for encoder models, DAR…
-
LLM-guided framework enhances neural architecture search proxies
Researchers have developed Bi-EZP, a novel bilevel framework designed to improve the discovery of ensemble zero-cost proxies for neural architecture search (NAS). This framework separates the discrete structural optimiz…
-
LP-NAS framework uses linear programming for efficient neural architecture search
Researchers have introduced LP-NAS, a novel framework for Neural Architecture Search (NAS) that leverages linear programming principles. This method aims to automate the design of neural network architectures by treatin…
-
New NAS methods optimize neural networks for compactness and performance
Researchers have developed a new framework for Neural Architecture Search (NAS) that uses continuous relaxations to optimize neural network architectures more efficiently. This approach, which includes methods like NAS-…
-
Darts library unifies foundation models for zero-shot time series forecasting
A new collection of foundation models for time series forecasting has been developed within the Darts Python library. This initiative aims to unify the interfaces of various pre-trained models, including Chronos-2, Time…
-
Planetary rover ERNEST uses AI for all-terrain locomotion
Researchers have developed a novel control system for a planetary rover named ERNEST, which features an actively articulated suspension. This system utilizes a single neural network controller trained via reinforcement …
-
Ukraine's FPV drones achieve 102km strike range
Ukrainian forces have developed FPV drones capable of striking targets up to 102 kilometers behind Russian lines, a significant advancement from their earlier capabilities. These drones, initially improvised by racing e…
-
New SWAP-Score metric evaluates neural networks without training
Researchers have introduced SWAP-Score, a novel zero-shot metric designed to evaluate neural networks without requiring training. This method measures a network's expressivity using sample-wise activation patterns and d…
-
DARTS method optimizes covariate acquisition for budget-constrained sequential experiments
Researchers have developed DARTS (Dynamic Adaptive Rerandomization via Thompson Sampling), a novel method for optimizing covariate acquisition in budget-constrained sequential experiments. This approach treats the proce…