With ChatGPT becoming the favourite application of millions, it is only plausible that more and more companies will come forward with their own iterations of ChatGPT. Chinese tech giant Huawei which is known for its innovative smartphones is gearing up to launch its own AI chatbot. As of now, ChatGPT, Bard, Ernie bot, and many other AI chatbots are engrossing people worldwide. Now, it seems ChatGPT may soon have a worthy rival.
According to the latest reports, Huawei is likely to launch its ChatGPT rival by July 7. The new AI software from Huawei is known as Pangu Chat and is named after the company’s large-scale learning model. Pangu Chat will be launched at the Huawei Developer Conference (Cloud) event.
What do we know about Pangu Chat?
Recent reports suggest that Pangu Chat will only be initially available for business and the company is reportedly planning to sell it to the government and enterprise customers. It is learned that Pangu Chat is based on the large-scale Pangu cloud model that comes with 100 billion parameters. This is lower than OpenAI’s GPT-3 which is said to have over 175 billion parameters and GPT-4 is rumored to have crossed 1 trillion parameters.
Parameters are values that define the skill of the model. Based on the unconfirmed reports, Pangu Chat with its 100 billion parameters seems to be no match to OpenAI’s GPT-4.
Reportedly, Huawei used more than 2000 Huawei Ascend 910 chipsets for over two months of data training capabilities. Industry analysts revealed that over 4000 GPU/TPU cards are in training large-scale model every year by the company. Interestingly, the computing power cost for Huawei has crossed over 960 million yuan in three years.
It has also been reported that based on tests of the Pangu model, the Pangu-E large model had an output of nearly 1.085 trillion parameters which is equivalent to GPT-4’s capability. The Pangu-E large-scale model is developed using the deep learning framework known as MindSpore. The framework can generate conversations at par with GPT-3.5.
More on Pangu Model
The model was launched by Huawei in 2021 and its iteration version 2.0 was unveiled in April last year. At present, the model is powered by a natural language processing (NLP) large model, meteorological model, and CV large model.
The primary layer of the Pangu Series AI Large Models includes NLP large models, scientific computing models, CV large models, etc. And the uppermost layer comes with the company’s large models developed with partners.
Reportedly, it is the first Chinese pre-trained large model with over 100 billion parameters. The model has the ability to learn from small samples, maps, text, audio, and three-modal large-scale learning.
It is to be noted that the company has not made any official confirmation and all the above-stated information has been compiled from various sources. If Huawei is actually planning to launch a powerful AI chatbot, then it would be yet another rival to ChatGPT. In March this year, Chinese search giant Baidu unveiled its Ernie Bot with reportedly 260 billion parameters.