Pathways Language Model | PaLM

In this video, I will talk about Google’s new large language model : Pathways Language Model and PaLM for Short. It is trained on 540 Billion parameters. For queries: You can comment in comment section or mail me at [email protected] This model can perform many natural language processing tasks: • Generating code from text • Answering Math word problem • Explaining a joke • common-sense reasoning, • translation, and many more Total dataset size of PaLM is 780 billion tokens. The training dataset of PaLM consist of a high-quality web documents, Social media conversations , books, GitHub code, Wikipedia, and many more with a “lossless” vocabulary. PaLM showed multiple breakthrough capabilities on different difficult tasks such as • language understanding and generation, • multi-step arithmetic code-related tasks, • common-sense reasoning, • translation, and many more Pathways Language Model (PaLM) uses a technique called chain of thought prompting to do complex tasks like solve math word problems