Tech open source API makes generative AI accessible | TechCrunch

Today, almost everyone is trying to get a piece of the generative AI action. While most of the focus remains on model vendors like OpenAI, Anthropic, and Cohere, or larger companies like Microsoft, Meta, Google, and Amazon, there is also the fact that many startups are trying to solve generative artificial intelligence in various ways. Questions of wisdom. This is such a new start-up company. While lacking the brand recognition of some other players, it has the largest open source model API with over 12,000 users per company. This open source appeal tends to attract investors, and the company has raised $25 million to date.

Qiao Lin, co-founder and CEO of Fireworks, noted that her company does not train basic models from scratch, but helps fine-tune other models to meet the specific needs of the business. “It can be an off-the-shelf, open-source model, or it can be a model that we can tune, or our customers can tune the model themselves. All three varieties can be served through our inference engine API,” Qiao told TechCrunch .

As an API, developers can plug it into their applications, train models of their choice on the data, and add generative artificial intelligence capabilities such as quick questions. Joe says it’s fast, efficient, and produces high-quality results.

Another advantage of the Firework approach is that it allows companies to try multiple models, which is important in a rapidly changing market. “The idea is that we want to enable users to iterate and try multiple models and have effective tools to inject their data into multiple models and test them with the product,” she said.

Perhaps more importantly, they lowered costs by limiting the model size to between 7 billion and 13 billion tokens, compared to over 1 trillion tokens in ChatGPT4. While this limits the range of words that large language models can understand, it enables developers to focus on smaller, more focused sets of data designed to address more limited business use cases.

Qiao is uniquely qualified to build such a system, having previously worked at Meta, leading the AI ​​platform development team with the goal of building a rapid, scalable development engine to power AI for all Meta products and services. This knowledge came from working at Meta and creating an API-based tool that allows any company to have this capability without having the level of engineering resources that a company the size of Meta would have.

The company raised $25 million in 2022, led by Benchmark, with participation from Sequoia Capital and angel investors including Databricks and Snowflake. The latter two are particularly interesting strategic investors because they are both data storage tools, and Fireworks will enable users to store this information at work.

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button