DeepL’s Model That Translates Better Than Google Translate

Since 2017, it has become one of Google Translate’s toughest competitors in machine translation. DeepLhas now come up with a brand new large language model (LLM). It has been stated that the new model of the Germany-based initiative is better than many translation systems today.

The new language model is specifically focus on translation and editing It is described as a model developed for.

It is said to outperform Google Translate and ChatGPT

According to the company’s statements, the purpose of the model is, produce more human-like translations and reduce the risk of hallucinationsWe know that hallucination is a huge problem in AI models. This is when models produce outputs that are not real or are false.

It was also added that the model, unlike its competitors, is not connected to the internet for training. Instead, DeepL own data also uses language experts It was stated that it trained the model to improve translation quality.

According to the company’s claim, the model Google, OpenAI and Microsoft It performs better than leading companies in translation today, such as Google Translate. In other words, it is better than Google Translate, which almost everyone uses, and ChatGPT, which has improved itself a lot recently. In tests, Google Translate has been shown to reach the quality of DeepL’s model. 2 times more to regulation, ChatGPT’s 3 times more more than that was observed to be needed.

In addition, tests were conducted among their own models. The results were in English, Japanese and Chinese. 1.7x improvementin the English-German pairing 1.4x improvement It was observed that.

The system that includes the new model is currently only those who subscribe to the paid subscription system DeepL Pro for English, German, Japanese and Chinese It was offered in four languages. Users can use the new generation model by selecting it from the web. DeepL Pro fees start from $8.74 per month in our country.