China develops generative AI that beats ChatGPT and generates 18 words per second

In a new development in the sphere of artificial intelligence, the Chinese company SenseTime has unveiled the model SenseNova 5.0. The AI ​​model appears to have outperformed Generative Pre-trained Transformer 4 (GPT-4), OpenAI’s large multimodal language model.

Medium reported that SenseNova 5.0 managed to outperform the acclaimed GPT-4 on a number of performance standards. These measures include logical reasoning and creative writing, as well as the ability to generate images.

The new model showed improved ability to understand and generate human-like text, demonstrating practicality and effective solutions applied to real-world applications. His ability to process words could allow him to manage the external relations of entire companies.

SenseNova: a hybrid model

SenseNova 5.0, SenseTime’s largest model, was unveiled on April 8, 2024, at a Tech Day event in Shanghai. They also launched the “Cloud-To-Edge” full-stack large product matrix.

This new model of generative AI represents a significant advancement in the realm of AI. The model works as a hybrid, integrating both transformer and recurrent neural network architectures. Additionally, it was trained on a diverse dataset of over 10 billion tokens from multiple languages ​​and sources.

PR Newswire reported that SenseNova 5.0 underwent training on more than 10 TB of tokens, covering a large amount of synthetic data.

His main gains are in knowledge, mathematics, reasoning and coding skills. The project also relied on a set of experts to complete the training sessions.

SenseNova 5.0’s superior performance rivals ChatGPT-4 with its advanced learning optimization techniques and can effectively handle large volumes of data. Because of this, it is capable of producing more accurate results and providing applications in different industries.

‘Constant renewal, daily renewal and further renewal’.

“In the era of Generative Artificial Intelligence (AGI), the three elements of data, algorithms and computing power are undergoing a new evolution,” Dr. Xu Li, President and CEO of SenseTime, said in an official statement . “The number of model parameters will increase exponentially and the volume of data will grow massively with the introduction of multiple modalities, leading to a continued surge in demand for computing power.”

The inference speed of SenseNova Edge-side Large Language Model has achieved industry-leading performance. It can generate 18.3 words per second on mid-range platforms and a whopping 78.3 words per second on flagship platforms.

The diffusion model also achieved the fastest inference speed in the industry. The inference speed of LDM-AI edge-side image diffusion technology takes less than 1.5 seconds on a mainstream platform and supports the output of high-definition images with a resolution of 12 million pixels and above, as well as functions of image editing such as proportional expansion, freeform and image rotation.

The SenseTime Integrated Large Model (Enterprise) edge device was developed in response to the growing demand for AI from key industries such as finance, coding, healthcare and government services. Compared to other similar products, the device performs accelerated searches with only 50 percent CPU utilization and reduces inference costs by approximately 80 percent.