853: Generative AI for Business — with Kirill Eremenko and Hadelin de Ponteves

Topics covered
Popular Clips
Questions from this episode
- Asked by 124 people
- Asked by 56 people
- Asked by 42 people
- Asked by 41 people
- Asked by 35 people
- Asked by 27 people
- Asked by 27 people
- Asked by 24 people
- Asked by 23 people
- Asked by 20 people
Episode Highlights
Lifecycle
Foundation models are the backbone of many AI applications, offering a structured approach from inception to deployment. explains that these models undergo an eight-step lifecycle, starting with data preparation and model architecture selection, followed by the intensive pre-training phase 1. This process, while costly and resource-intensive, is crucial for developing models like ChatGPT, which serve as a foundation for further applications 2. highlights the accessibility of these models through platforms like AWS Bedrock, which simplifies their application in various domains 3.
Foundation models, once developed, can be used as a basis to build your own applications.
---
These models are not only foundational but also adaptable, allowing businesses to leverage them for specific needs.
Customization
Customizing foundation models is essential for aligning them with specific business requirements. discusses the importance of fine-tuning, which involves adjusting models during training to better fit particular use cases 4. This process can be cost-effective, especially when using techniques like domain-specific fine-tuning or instruction-based fine-tuning, as shares from his experience 5.
Fine-tuning is like a type of modification during training, closer than to inference.
---
These modifications ensure that the models deliver precise and relevant outputs, enhancing their utility in business applications.
Selection
Selecting the right foundation model involves evaluating multiple criteria, including cost, performance, and environmental impact. emphasizes the importance of considering long-term costs and the potential scalability of models 6. adds that compliance, licensing, and environmental considerations are also crucial factors in the decision-making process 7.
There are 12 considerations that businesses tend to look out for when selecting a foundation model.
---
By understanding these factors, businesses can make informed decisions that align with their strategic goals and operational needs.
Related Episodes


743: How to Integrate Generative AI Into Your Business — with Piotr Grudzień
Answers 383 questions

687: Generative Deep Learning — with David Foster
Answers 383 questions

679: The A.I. and Machine Learning Landscape — with investor George Mathew
Answers 383 questions

847: AI Engineering 101 — with Ed Donner
Answers 383 questions

671: Cloud Machine Learning — with Kirill Eremenko and Hadelin de Ponteves
Answers 383 questions

783: Generative AI for Solar Power Installation — with Navdeep Martin
Answers 383 questions

735: AI Product Management — with Google DeepMind's Head of Product, Mehdi Ghissassi
Answers 383 questions

763: The Best AI Startup Opportunities — with venture capitalist Rudina Seseri
Answers 383 questions

649: Introduction to Machine Learning — with Kirill Eremenko and Hadelin de Ponteves
Answers 383 questions

841: AI Vision, Agents and Business Value — with Andrew Ng
Answers 383 questions

842: Flexible AI Deployments Are Critical — with Chris Bennett and Joseph Balsamo
Answers 383 questions

711: Image, Video and 3D-Model Generation from Natural Language — with Dr. Ajay Jain
Answers 383 questions

777: Generative AI in Practice — with Bernard Marr (@bernardmarr)
Answers 383 questions














