AI has become very popular – especially text-generating AI, also known as large language models (think ChatGPT-style models). In a recent one reconnaissance Of the nearly 1,000 institutions, 67.2% say they see adoption of large language models (LLMs) as a top priority by early 2024.
But barriers stand in the way. According to the same survey, a lack of customization and flexibility, coupled with an inability to maintain company knowledge and intellectual property, was—and still is—preventing many companies from deploying LLMs in production.
This got Varun Phummadi and Esha Mandeep Dean thinking: What could the solution to the enterprise LLM accreditation challenge look like? In search of one, they founded GB mla startup that is building a platform that allows companies to deploy LLMs in-house — ostensibly cutting costs and preserving privacy in the process.
“Data privacy and customization of LLMs are two of the biggest challenges organizations face when adopting LLMs to solve problems,” Fumadi told TechCrunch in an email interview. “Giga ML addresses both of these challenges.”
Giga ML offers its own set of LLM certifications, the “X1 Series,” for tasks like creating code and answering common customer questions (for example, “When can I expect my order to arrive?”). The startup claims that the models, built on Meta’s Llama 2, outperform popular LLMs in certain benchmarks, notably MT seat Test suite for dialog boxes. But it’s hard to say how the X1 compares in terms of quality; This reporter tried Giga ML Online demo But I encountered technical problems. (The application timed out regardless of which prompt you typed.)
Even if Giga ML models We are Although they excel in some aspects, can they really make a splash in the ocean of open source and offline LLM programs?
Speaking to Vummadi, I got the sense that Giga ML is not so much trying to create the best performing LLM software but is instead building tools that allow companies to fine-tune LLMs locally without having to rely on third-party resources and platforms.
“Giga ML’s mission is to help organizations securely and efficiently deploy LLMs on their on-premises infrastructure or virtual private cloud,” said Fumadi. “Giga ML simplifies the process of training, fine-tuning and running LLM courses by taking care of it through an easy-to-use API, removing any associated hassles.”
Fomade stressed the privacy benefits of running forms offline, which are likely to be compelling to some companies.
Predibase, a low-code AI development platform, found that less than a quarter of organizations are comfortable using commercial LLM certifications due to concerns about sharing sensitive or private data with vendors. Nearly 77% of survey respondents said they either do not use or do not plan to use commercial LLMs beyond prototyping in production – citing issues with privacy, cost, and lack of customization.
“C-suite level IT managers find Giga ML offerings valuable due to secure on-premise deployment of LLMs, customizable models tailored to the specific use case and fast inference, ensuring data compliance and maximum efficiency.” Fumade said.
Giga ML, which has raised approximately $3.74 million in venture capital funding to date from Nexus Venture Partners, Y Combinator, Liquid 2 Ventures, 8vdx and several other companies, plans in the near term to grow its two-person team and increase product R&D . A portion of the capital will be allocated to support Giga ML’s customer base as well, Which currently includes unnamed “institutional” companies in finance and healthcare, Fumade said.