Building on its collaboration with Meta on open innovation for AI, IBM plans to host Meta’s Llama 2-chat 70 billion parameter model in the watsonx.ai studio – with early access now available to select clients and partners.
The move strengthens the relationship between Meta and IBM which already includes work with open source projects such as the Meta-developed PyTorch machine learning framework and the Presto query engine used in watsonx.data.
It will also support IBM’s strategy of offering both third-party and its own AI models. Currently in watsonx.ai, AI builders can leverage models from IBM and the Hugging Face community, which are pre-trained to support a range of Natural Language Processing (NLP) tasks including question answering, content generation and summarization, text classification and extraction.
The future availability of Llama 2 in watsonx.ai will be another milestone on IBM’s generative AI roadmap and is expected to be followed by the coming releases of its AI Tuning Studio, additional AI models in watsonx.ai, and FactSheets in watsonx.governance.
IBM says it is committed to keeping trust and security principles at the forefront as it continues to roll out its generative AI capabilities.
For instance, the company says, when users run the Llama 2 model through the prompt lab in watsonx.ai, they can toggle on the AI guardrails function to help automatically remove harmful language from the input prompt text as well as the output generated by the model. Meta also provides an account of their fine-tuning methodology used in their large language models.
Furthermore, IBM Consulting has the expertise of 21 000 data, AI and automation consultants in addition to its Centre of Excellence for Generative AI comprised of more than 1 000 consultants with specialised generative AI expertise. These experts can work with clients to help tune and operationalise models for targeted use cases aligned to their specific business requirements.