30B-A3B
PulseAugur coverage of 30B-A3B — every cluster mentioning 30B-A3B across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
New Multiscale Policy Optimization enhances LLM agent performance
Researchers have developed a new method called Multiscale Policy Optimization (MPO) to improve the performance of long-horizon Large Language Model (LLM) agents. MPO addresses the challenge of assigning credit for actio…
-
NVIDIA releases SEA-LION-v4.8 models for Southeast Asian languages
NVIDIA has released a technical report detailing the SEA-LION-v4.8 family of models, built upon the NVIDIA Nemotron 3 architecture. These models, including 30B-A3B and 120B-A12B variants, have been enhanced with Southea…
-
OpenResearcher pipeline enables offline synthesis of AI research trajectories
Researchers have developed OpenResearcher, an open-source pipeline designed for synthesizing long-horizon research trajectories for training deep research agents. This pipeline operates offline, utilizing three explicit…
-
New European open-source model Soofi S released for local use
A new open-source large language model named Soofi S, with a 30 billion parameter version labeled 30B-A3B, has been released. This European-developed model is designed for local execution and is in its early stages of d…
-
Loopie Transformer architecture achieves gold-medal performance on Olympiads
Researchers have introduced Loopie, a novel looped Transformer architecture designed to overcome previous limitations in scaling. The Loopie series includes two Mixture-of-Experts (MoE) models: a 20B-parameter version w…