Google Releases Big Language Model PaLM 2, Ready to Compete with GPT-4?

Liputan6.com, Jakarta – Google released PaLM 2, Google’s next generation Large Language Model / LLM (large language model) which will compete with other large language models such as GPT-4 (the model behind ChatGPT) from OpenAI.

PaLM 2 is an advanced language model with enhanced multilingual, reasoning, and coding capabilities.

Multilingual

PaLM 2 has undergone an intensive training process on multilingual texts, covering more than 100 languages. According to Google, this training process has significantly improved his ability to understand, produce, and translate text with certain nuances, including idioms, poetry, and riddles, across a wide spectrum of languages. In addition, PaLM 2 goes beyond the advanced language proficiency exam at the “mastery” level.

reasoning

The diverse PaLM 2 dataset includes scientific papers and web pages containing mathematical expressions. As a result, the model shows increased proficiency in logic, common sense reasoning, and mathematics.

Coding

PaLM 2 is pre-trained with a large number of publicly available source code datasets. This specialized training allows models to excel in popular programming languages ​​such as Python and JavaScript, while also generating specialized code in languages ​​such as Prolog, Fortran, and Verilog.

Despite its various upgrades, Google says that PaLM 2 is faster and more efficient than its predecessor. Plus, it comes in a variety of sizes, making it adaptable for different use cases.

Google continues to improve their artificial intelligence (AI) capabilities. One of the uses of AI is to be able to provide better answers when internet users search on Google Search.