Base Labs releases top-ranked decision models mammoth-1 and and elephant-1 (www.baseten.co)

Base Labs has unveiled two top-ranking decision models, mammoth-1 and elephant-1, which excel in categorizing and scoring tasks according to the Jev Decision Index. Mammoth-1 leads among all tested decision models, while elephant-1 stands out as the top 9B decision model, both available for free trials via Baseten Model APIs with a commitment to zero data retention. These models are designed to streamline decision-making by providing probability distributions for various types of questions—choices, scores, and boolean queries—without the computational overhead associated with generative language models. The introduction of mammoth-1 and elephant-1 is significant for the AI/ML community as they fill a crucial niche in decision-making tasks, offering improved speed and cost-effectiveness over traditional methods. By focusing solely on decision outputs and employing customizable confidence thresholds, these models enhance efficiency and reliability in areas such as ticket triage, spam detection, and safety checks. Trained on a comprehensive dataset of 6.21 million examples from multiple domains, the models demonstrate robust capabilities while greatly reducing response times compared to generative models, achieving up to a tenfold speed improvement.