Galileo’s Luna redefines GenAI diagnostics, boasting 97% lower costs and 11x faster speeds

GalileoA trailblazer in enterprise generative AI has been unveiled. Galileo LunaA groundbreaking suite of Evaluation Foundation Models (EFMs) that promises to transform how enterprises evaluate their GenAI systems. With Luna, Galileo aims to address the critical challenges of speed, cost, and accuracy that hinder the widespread adoption of generative AI in production environments.

“Galileo built Luna to address the limitations of existing GenAI diagnostic methods, which were slow, expensive and often inaccurate,” said Galileo co-founder and CEO Vikram Chatterjee in an interview with VentureBeat. “This motivation stems from the need for ultra-low-latency, cost-effective, and high-accuracy diagnostics in production environments.”

Luna’s development marks an important milestone for Galileo, which has been at the forefront of the GenAI enterprise since its founding in early 2021. For Luna’s creation.

Luna, Galileo’s groundbreaking suite of evaluation foundation models, outperforms AI evaluation methods in benchmark comparisons of area under the receiver operating characteristic curve (AUROC) scores. The high AUROC values, which reach 0.78, demonstrate Luna’s superior accuracy in assessing enterprise generative AI systems, outperforming competitors such as GPT-3.5, Trulens Groundedness, and RAGAS Faithfulness. (Image credit: Galileo)

Purpose-built models redefine speed, cost and accuracy.

At the heart of Luna’s innovation are its purpose-built miniature language models, which are carefully designed for specific diagnostic tasks such as deception detection, context quality assessment, data leakage prevention, etc. Tham, and malicious quick identification. This unique design allows Luna to deliver unparalleled performance in three key metrics: speed, cost, and accuracy.

“Luna outperformed GPT-3.5 in speed, cost and accuracy through several innovations,” explained Chatterjee. “Luna uses small, purpose-built language models that are designed for specific evaluation tasks, significantly reducing computational overhead and cost. This design choice allows these evaluations to are 97% cheaper and 11x faster than GPT-3.5.

But it’s not just about speed and cost. Luna also boasts industry-leading accuracy, outperforming previous methods by up to 20 percent in detecting fraud, instant injection, personally identifiable information (PII) and more. “Advanced techniques such as multi-head miniature language models and intelligent chunking ensure that Luna models retain context better and provide more accurate diagnostics,” added Chatterjee.

Compared to the monthly costs of evaluating 1 million queries, Galileo’s Luna significantly undercuts other methods, costing only $175 per month. Luna’s purpose-built miniaturized language models enable extremely low-cost evaluation, making it 97 percent more cost-effective than alternatives such as GPT-3.5 at $6,248 per month, RAGAS Faithfulness at $7,994 per month, and Trulens Groundedness. $16,641 per month. . (Image credit: Galileo)

A revolution in assessment without ground truth datasets

One of the most notable aspects of Luna is its ability to operate without the need for traditional ground truth datasets. By leveraging diagnostic models pre-trained on diverse, domain-specific datasets, Luna eliminates the time-consuming and expensive process of creating custom test sets. This innovation streamlines the assessment process and reduces reliance on extensive human-generated data.

Luna’s potential applications are vast, with Chatterjee highlighting its relevance in industries that demand high reliability and speed in AI diagnostics. “Luna is particularly powerful in large-scale enterprise applications where volume and throughput are critical (i.e. millions of queries per month). We’re seeing Fortune 100 enterprises in healthcare, finance and telecom use Luna in particular. But consider it effective,” he said.

Galileo’s Luna delivers unprecedented speed in AI diagnostics, with a latency of just 0.232 seconds to process a single query. This indicates a significant improvement over other methods, such as GPT-3.5 at 2.5 seconds, Galileo ChainPool at 3.0 seconds, Troll’s Groundiness at 3.4 seconds, and RAGAS Fidelity at 5.4 seconds. Luna’s purpose-built miniaturized language models enable ultra-low-latency evaluation, making it 11 times faster than competitive approaches. (Image credit: Galileo)

Customization and continuous evolution given the rapid development of GenAI

Use cases range from real-time monitoring of AI outputs to detecting deception in AI-generated content to ensure the safety and quality of chatbot interactions. And with Galileo’s Fine Tune product, Luna can be customized to meet specific customer needs, achieving accuracy of 95% or more for critical tasks in industries such as pharmaceuticals and financial services. level can be achieved.

As the creative AI landscape continues to rapidly evolve, Galileo is committed to being at the forefront of innovation. Chatterjee emphasized that Luna will scale in three key ways: expanding support for more types of diagnostic work, continuously improving accuracy, and further reducing costs and delays.

“Galileo is committed to pushing the limits of what is possible in AI diagnostics and helping enterprises bring reliable AI to production,” said Chatterjee. “As the creative AI landscape continues to evolve, Galileo remains dedicated to providing its clients with advanced diagnostic capabilities that put AI into action for businesses to foster trust and confidence among consumers. “

With the launch of Luna, Galileo has solidified its position as a leader in enterprise GenAI diagnostics. As more organizations look to harness the power of generative AI, Luna’s ability to deliver fast, low-cost, and accurate assessments will be a key factor in driving widespread adoption and unlocking the full potential of this transformative technology. It will prove.

Credit : venturebeat.com