Ever since OpenAI released its GPT-4 LLM model, AI and research organizations around the world have been trying to match or outperform the performance of GPT4 model. So far, some of them have reached the performance of GPT-3.5, but GPT-4 remained the one to beat.
Today, Baidu revealed their latest LLM model named ERNIE 3.5. Baidu claims that this ERNIE 3.5 model surpasses GPT-3.5 model in comprehensive ability scores and outperforms GPT-4 in several Chinese language capabilities.
Through the power of ERNIE 3.5, the ERNIE Bot now offers improved creative writing, Q&A, reasoning, and code generation.
Baidu is also following OpenAI with the concept of plugins. Similar to ChatGPT plugins, ERNIE’s plugins will allow you to expand the capabilities of the bot.
The default built-in plugin “Baidu Search” enables ERNIE Bot with the ability to generate real-time and precise information. The ChatFile plugin enables long text summary and Q&A. ERNIE Bot will add more high-quality plugins in the future.
In ERNIE 3.5, we’ve implemented cutting-edge strategies from PaddlePaddle, including adaptive hybrid parallel training technology and mixed-precision computing, Dr. Wang explained. These enhancements, combined with optimized data sources and data distribution, have accelerated our model’s iteration speed, bolstered its efficacy, and ensured its safety.