Alibaba releases two open-sourced AI fashions to rival Meta’s Llama 2

by Jeremy

Alibaba Group, the Chinese language tech and e-commerce big, introduced the discharge of two open-sourced synthetic intelligence (AI) fashions from its cloud computing division on Aug. 3, in response to a press launch. 

Its two giant language fashions (LLMs) are dubbed Qwen-7B and Qwen-7B-Chat, with every mannequin having 7 billion parameters. Alibaba stated these two fashions are small-size variations of the Tongyi Qiawen, which the corporate launched in April.

These new fashions are geared toward serving to introduce AI within the operations of small to medium-sized companies.

The corporate stated Qwen-7B and Qwen-7B-Chat have varied capabilities that may be interesting to enterprises resembling having the ability to “code, mannequin weights, and documentation shall be freely accessible to lecturers, researchers and business establishments worldwide.”

Alibaba’s newest LLMs are additionally the primary launched from a Chinese language tech firm to be open-sourced. Nonetheless, it stated companies with over 100 million month-to-month lively customers will want a license.

On Aug 1., the corporate additionally introduced an replace within the type of a vector engine to its AnalyticDB knowledge warehousing service, which is able to permit its company shoppers to shortly create customized generative AI purposes.

Associated: AI chip developer will get $100 million from Samsung and Hyundai

This growth comes after Meta’s launch of its open-sourced LLM named Llama 2, which it launched with Microsoft on July 16.

Meta says its Llama 2 was educated utilizing 40% extra public knowledge and in a position to course of twice as a lot context in comparison with its predecessor. It is usually open-sourced and the most important model of Llama 2 options 70 billion parameters. 

Just like Alibaba’s newest it additionally requires a license from company customers with over 700 million customers.

On July 26, Alibaba introduced what it referred to as “the primary coaching and deployment resolution for your complete Llama2 collection in China” after it deployed a Llama 2 resolution for companies, with a purpose to develop AI-powered software program and instruments.

Journal: Make 500% from ChatGPT inventory suggestions? Bard leans left, $100M AI memecoin: AI Eye