In addition to the Granite models, IBM is enhancing its Watsonx.data platform with generative AI capabilities, allowing users to explore, enrich, display, and improve data for AI applications via a chatbot-like interface. IBM is also introducing Watsonx.governance, a toolkit designed to protect client information, spot bias and drift in models, and ensure adherence to ethical norms. These developments are part of IBM's efforts to strengthen its position in the AI industry and provide relevant AI solutions for organizations across various sectors.
Key takeaways:
- IBM has announced the addition of new generative AI models and capabilities to support its Watsonx data science platform. The models, known as the Granite series, are comparable to large language models like OpenAI's GPT-4 and ChatGPT.
- The Granite models, developed by IBM Research, use a "Decoder" architecture and can process 13 billion parameters with a single V100-32GB GPU. They can perform tasks like summarization, question-answering, categorization, content creation, insight extraction, and named entity identification.
- IBM is also enhancing its Watsonx.data platform with generative AI capabilities and a vector database feature for retrieval-augmented generation (RAG). Additionally, in response to concerns about consumer privacy and ethical AI practices, IBM is introducing Watsonx.governance.
- The company trained these models using specialized datasets totaling 1 trillion tokens from 7 terabytes of data. The datasets cover a range of subjects, ensuring the models are knowledgeable in language and vocabulary specific to various sectors.