Deploying open-source LLMs, such as the MPT and Llama models, is filled with challenges. The process can often be difficult, with the need to fine-tune for optimal performance or navigate intricate hyperparameters. Not to mention the issue of privacy that comes with external servers like OpenAI. To combat these difficulties and provide a seamless experience; large language model (LLM) provider Giga ML has secured $3.6 million in seed funding round led by Nexus Venture Partners.
The round was also joined by Y Combinator, Liquid 2 Venture, and 8vdx. It also saw participation from prominent angels like Garry Tan (Y Combinator president and CEO).
Enterprises are looking to deploy AI and LLM-powered solutions for multiple internal and external use cases. Though data security and compliance pose difficult challenges, open-source models have emerged as strong contenders for enterprise adoption.
Giga ML is the solution for enterprises to customise and deploy large language models in a secure way. The platform allows businesses to deploy LLMs as robust as GPT-4 directly on their own servers. This eliminates the need to send sensitive information to external servers like those of OpenAI.
“We are excited to partner with the Giga ML team in their journey to securely enable enterprise adoption of LLMs; we have been big believers in the open-source ecosystem, and we think this will play out equally or even more strongly in the case of LLMs and enterprise adoption. However, there are significant challenges today around the lack of know-how of internal teams to put these models in production in a secure way. Giga ML has taken up on the challenge to become the platform for LLM deployments in enterprises,” Abhishek Sharma, Managing Director, Nexus Venture Partners.
More about Giga ML
GigaML was founded three months ago by IIT-Kharagpur alumni Varun Vummadi and Esha Manideep Dinne. The platform helps enterprises secure on-premise deployment of large language models (LLMs).
Giga ML’s X1 Large 32k model is an advanced iteration of the llama2 70B 4k model, designed for enhanced performance. It surpasses competitors like Mosaic ML and Together AI in fully fine-tuning a 32k context length. The company has reached technical milestones through fine-tuning Llama 2.
“At Giga ML, our utmost priority is to ensure the privacy and security of enterprise data. Our goal is to position our company as the leading provider of On-Premise Secure LLMs, allowing enterprises to leverage the power of LLMs without compromising their data integrity,” said co-founder Varun Vummadi.
Already, industries like healthcare, legal, and finance are exploring the potential of Giga ML’s solutions for various applications. This leverages GigaMl’s offerings for customer support, internal knowledge search, and code generation, to enhance engineering team productivity.
Just 15 days after its public launch the company’s website has rapidly risen to become the fourth most-visited among startups in Y Combinator’s S23 batch.
“GigaML is the missing layer that Llama 2 really needs – open source LLM models are a keystone to the future of AI we want to live in, free of lock-in to big tech or those beholden to big tech,” shared Garry Tan, CEO and President of Y Combinator.
Earlier this year, Nexus closed its fund VII at $700 million to back AI, SaaS, fintech, and commerce startups. Its portfolio includes companies like Postman, Apollo.io, MinIO, Fingerprint, Nx, Druva, Hasura, Observe.ai, Delhivery, H2O.ai, Quizizz, Sibros, TileDB, Turtlemint, Unacademy, Zepto, and Zomato, among others.