Michael Fauscette’s Post

Continuing on the Generative AI common questions theme, the second most frequent question: "How can we evaluate models to determine the best one?" This has been challenging in AI/ML in general and moreso in Gen AI because of the nature of foundation models and the vast array of workloads they can support. I am excited to share some of the work on operationalizing LLM Evaluation that launched this weekend. Blog: https://lnkd.in/gpyXkh7G Repo: https://lnkd.in/gRcX_3W4

@_odsc In this article, we briefly discuss the various phases of IDP and how generative AI is being utilized to augment existing IDP workloads or develop new IDP workloads. #DataScience #AI #ArtificialIntelligence https://lnkd.in/ekJfRwBQ

In this article, we briefly discuss the various phases of IDP and how generative AI is being utilized to augment existing IDP workloads or develop new IDP workloads. #DataScience #AI #ArtificialIntelligence https://hubs.li/Q024C_kw0

The evolution and capabilities of AI are fascinating! https://lnkd.in/eU6DWyme

With the rise of large language models (LLMs) what role will Edge AI play? Get the answer to this and other questions on the impact of Edge AI in this SNIA Q&A blog: https://lnkd.in/ewuk9xsN #AI #EdgeAI

Learn how to automate caption creation & search for images at enterprise scale using generative AI and Amazon Kendra. Read the blog 👉 https://go.aws/3KsTleS #AmazonKendra #GenerativeAI #MachineLearning

The introduction of generative AI provides another opportunity for Thomson Reuters to work with customers and once again advance how they do their work, helping professionals draw insights and automate workflows, enabling them to focus their time where it matters most. #aws #genai #ai #machinelearning https://lnkd.in/gajBT3hF

Are you aware of the risks that Generative AI can pose to your organisation's reputation and customer trust? It's a crucial topic that needs our attention. This insightful article highlights the challenges of detecting and managing risks such as toxicity, bias, and hallucinations in AI models. The good news is, Amazon SageMaker Clarify now offers FM evaluations, providing actionable insights and verifiable evaluation scores across various tasks. This new feature integrates with Machine Learning and Operation workflows, automating and scaling the ML lifecycle. In essence, this tool allows us to evaluate large language models for quality and responsibility, improving transparency and trust in language models on AWS. Read the full article for a comprehensive understanding. Happy Friday, everyone! Casey Jones #AI #MachineLearning #RiskManagement #AmazonSageMaker

I continue to maintain generative AI is the single most impactful technology innovation of our time.

Researchers from AWS AI Labs and USC Propose DeAL: A Machine Learning Framework that Allows the User to Customize Reward Functions and Enables Decoding-Time Alignment of LLMs Quick read: https://lnkd.in/gVWdReN6 Researchers have introduced a novel framework, DeAL (Decoding-time Alignment for Large Language Models), that reimagines the approach to model alignment by allowing for the customization of reward functions at the decoding stage rather than during training. This innovation provides a more flexible and dynamic method for aligning model outputs with specific user objectives. Paper: https://lnkd.in/gWB9fiFv #artificialintelligence #datascience #ai University of Southern California Amazon Web Services (AWS)