On 10 May 2023, Google’s CEO Mr. Sundar Pichai introduced the PaLM2, which is next generation LLM (large language model) and has capacity to support more than 100 languages. The biggest feature of PaLM2 is that it can do different tasks very easily. It is the smallest and fastest large language model till date. At the same time, they are also the most efficient. Google PaLM2 LLM (Large Language Model) can do coding, mathematics tests and creative & analysis very easily. The primary core of PaLM2 is its new TPU (Tensor Processing Units) infrastructure. TPU handles the calculations for such language models.
What’s New in Google PaLM2 LLM
Google PaLM2 has been heavily trained on multiple texts covering 100+ languages. It has also improved the ability to understand, write and translate short texts, puzzles and poetries in different languages.
It has also outperformed its previous model in mathematical and common-sense reasoning. According to Google, PaLM2 has been trained on a large amount of math and science texts.
What Google PaLM2 Large Language Model Can Do
- It has capability to interpret, generate and debug code for more than 20+ programming languages.
- PaLM2 is expertise in Python and JavaScript type languages.
- It can also generate code in Prolog, Fortran, and Verilog very easily.
- It can also help in providing documentation in any language for codes to developer.
Google has invested heavily in PaLM2 to make it smaller, faster and more efficient. Google has created a family of models according to size and performance. An example of this is the Gecko (a model belonging to a family of models made by Google) which is very compact and can run on a smartphone without an internet connection. Similarly, Bison, Otter and Unicorn are other models according to size and performance. They are also the biggest and most powerful.
Other Google Products Who Integrating PaLM2
Many Google teams are already using PaLM2. Google has introduced it much later. PaLM2 already providing assistance to 25 products and their features. The following products are supported by PaLM2.
- Google Workspace Apps like Gmail, Sheets and Docs) for writing purpose.
- Google AI Chatbot “Bard” for handling multiple languages and code writing.
- Med-PaLM 2 for medical applications. It is used for answering questions and providing summaries for various medical texts.
- Sec-PaLM for cybersecurity analysis. It analyses and explain malicious scripts behaviour.
- It is also powering MakerSuite which is famous for quickly building generative AI prototypes.
Conclusion
Google introduced PaLM2, the next generation large language model. It can support more than 100 languages. It is the smallest and fastest large language model ever developed. Google PaLM2 LLM is able to write code and mathematics tests very easily because of its primary core TPU.
Although Google announced it on 10th May 2023 but it is already available for developers (small group of developers) to use since March 2023.If any new developer wants to use it they can sign up and use it anytime from now.
Please comment and Contact Us for any discrepancy. Follow MambaPost on Facebook, Twitter and LinkedIn. For the latest Tech News checkout at MambaPost.com.
4 thoughts on “Google unveils its Powerful PaLM2 large language model”
Comments are closed.