My India First

My India First

Tencent, AI Mannequin: ‘Battle of hundred LLMs’: Tencent launches AI mannequin for enterprises

Tencent has introduced that its giant language mannequin (LLM), “Hunyuan”, is now obtainable for enterprises to make use of. The LLMs provided by expertise giants, corresponding to Google, Meta and OpenAI, are utilized by companies to develop ChatGPT-like synthetic intelligence (AI) chatbots.
The Chinese language tech big, which additionally owns WeChat, demoed the working of the mannequin, and mentioned Hunyuan had turn out to be the muse of greater than 50 of its services.
“By July, there are greater than 130 giant language fashions in China. A battle of 100 fashions has begun,” information company Reuters quoted Jiang Jie, Tencent’s vp, as saying.
Tencent mentioned its mannequin is able to conversing in each Chinese language and English and is “higher” than OpenAI’s ChatGPT in areas, which embody writing lengthy textual content with 1000’s of phrases and fixing sure math issues.
Tokens and parameters sport
Tencent highlighted that its LLM has greater than 100 billion parameters and was educated with greater than 2 trillion tokens. These two metrics are sometimes used to measure AI fashions’ energy and complexity.
Compared, OpenAI’s GPT-3 AI mannequin contained 175 billion parameters in 2020 and Meta Platform Inc’s Llama 2 mannequin had 70 billion parameters in 2023. Experiences prompt that Google’s PaLM 2 LLM that was unveiled at Google I/O, is educated on 3.6 trillion tokens and has 340 billion parameters.
Tencent claimed that its mannequin experiences 30% much less hallucination in comparison with Llama 2. Hallucination is an idea the place AI fashions generate incorrect data however current it as if it’s a reality.
Just lately, a number of Chinese language tech companies, together with Baidu Inc and SenseTime Group unveiled their very own AI fashions.



Source link