New AI models attempting to challenge the dominance of OpenAI’s renowned GPT 4.
Large language model (LLM) appear on a regular basis, with the most recent coming from a Chinese company called SenseTime.
Chinese Co. SenseTime
The company has unveiled its latest AI model, SenseNova 5.0, which claims to outperform GPT 4 in benchmarks.
Of course, real-world performance and benchmark results are not the same thing.
But SenseNova 5.0’s test results show that it outperforms OpenAI’s flagship AI in logical reasoning, creative writing, and other areas.
It is also more capable of understanding and producing human-like text than GPT 4, allowing it to apply effective solutions to real-world problems.
SenseNova 5.0
On April 8, 2024, at a Tech Day celebration in Shanghai, SenseTime unveiled its most expansive model to date, SenseNova 5.0.
This model combines the features of transformer and recurrent neural network architectures. It is also trained using a large and diverse dataset of over 10 billion tokens from various languages and sources.
According to PR, SenseNova 5.0 was trained using more than 10TB of tokens, including a significant amount of synthetic data.
Mixture of Experts
The model uses a ‘Mixture of Experts’ approach to manage a context window of approximately 200,000 during inference, improving its performance.
This context window is significantly larger than GPT 4 Turbo’s 128,000 inputs. However, let us not forget that being able to handle data effectively is a preferred performance metric over dealing with large amounts of data, which has yet to be proven.
Dr. Xu Li, the Chairman and CEO of SenseTime, stated:
In the era of AGI, the three elements of data, algorithms, and computing power are undergoing a new evolution.
The number of model parameters will increase exponentially, and the volume of data will grow massively with the introduction of multi modalities, leading to a continuous surge in demand for computing power.
To read our blog on “ChatGPT is headed to Nothing’s earbuds,” click here
