Alibaba introduced that its proprietary giant language model, a synthetic intelligence system known as Tongyi Qianwen, might be obtainable for public and enterprise entry all through China beginning Sept. 13.
Tongyi Qianwen is a ChatGPT-like giant language model educated on a corpus of English and Chinese textual content. While its actual specs aren’t recognized — early rumors indicated it will be educated with as many as 10 trillion parameters, 10 instances as many as OpenAI’s GPT4, however these stay unsubstantiated — Alibaba previously released two 7 billion-parameter open-source models primarily based on the Tongyi Qianwen structure.
Alibaba Cloud launched two open-source giant imaginative and prescient language fashions, Qwen-VL and Qwen-VL-Chat, that may comprehend photographs, texts, and bounding bins in prompts and facilitate multi-round query answering in each English and Chinese.
Learn extra: https://t.co/LrVNzNIRVy… pic.twitter.com/0t0Et6j1BU
— Alibaba Cloud Intelligence Group (@cloudintelgrp) August 30, 2023
Previously, Tongyi Qianwen had solely been obtainable to a restricted group of customers throughout its beta check section. The public rollout coincides with a latest loosening of restrictions associated to the use of synthetic intelligence (AI) applied sciences in the People’s Republic of China.
A set of pointers printed by the Chinese authorities in June dictated that, going ahead, all AI applied sciences launched to the public would require a particular vetting and certification course of.
The guidelines went into impact on Aug. 15. As Cointelegraph reported, a number of Chinese companies were given approval to launch fashions, together with Baidu, Tencent, TikTok and ByteDance.
Related: Tencent unveils ChatGPT rival in China amid continuing US AI chip ban
Among the provisions included in the up to date restrictions are guidelines barring the technology of photographs in the likeness of China’s president, Xi Jinping, and mandates indicating organizations will deal with objectionable content material inside a three-month interval. Previous variations of the laws known as for financial fines, however as Cointelegraph reported, those plans were axed.
As China explores a loosening of its laws, the United States has taken solely preliminary steps to control AI applied sciences. Most just lately, on Sept. 13, Senate Minority Leader Chuck Schumer hosted a litany of prime U.S. tech CEOs and founders in the primary of 9 scheduled boards to debate potential coverage concepts.