Artificial intelligence has become one of the most discussed topics since 2022. Although some experts consider 2023 the year of 'artificial intelligence,' there is still a long way to go. Following this, when we talk about artificial intelligence, the language models that form their backbone come to the fore.
The newly announced GPT-4, which forms the basis of ChatGPT, largely meets user requests. However, there are also artificial intelligence models in different fields.
Vectara, which prepared a list by looking at accuracy rates, turned this into a list by taking the ability of artificial intelligence chatbots to avoid hallucinations as a criterion.
Here are the best artificial intelligence models according to the prepared list:
- GPT 4
- GPT 3.5
- Llama 2 70B
- Llama 2 7B
- Llama 2 13B
- Cohere-Chat
- Cohere
- Anthropic Calude 2
- Mistral 7B
- Google Palm
- Google Plam-Chat
What is referred to here as hallucination is how accurately facts are conveyed. In other words, some artificial intelligence models sometimes distort facts in order to answer the questions asked. According to the tests conducted, it was revealed that GPT-4 has the lowest hallucination rate.
In addition to GPT-4, developed by OpenAI, GPT 3.5, which also comes from the same company, is in second place.
Google needs to make more progress as it is at the bottom of the list.
News Source: 12punto
