Amin Ahmad on how custom foundation models are revolutionizing how we search for information.
Subscribe: Apple • Spotify • Stitcher • Google • AntennaPod • Podcast Addict • Amazon • RSS.
Amin Ahmad, the co-founder of Vectara, has played a crucial role in developing a powerful API platform specifically tailored for developers. Vectara’s primary objective is to streamline the process of crafting conversational experiences. Equipped with state-of-the-art features like Retrieval, Summarization, and Grounded Generation, this platform guarantees exceptional performance and minimizes the likelihood of hallucinations in AI-generated content. Our conversation delves into the contemporary landscape of search and information retrieval, large language and foundation models, vector databases, and various other pertinent subjects.
Interview highlights – key sections from the video version:
- Hybrid search prior to the release of ChatGPT: BM25 + Neural Information Retrieval
- UX: generating & presenting results of hybrid search systems
- Infrastructure for modern search systems
- Vector Databases
- Embedding long documents
- Multimodal search
- Neural information retrieval pipelines
- A new notion of “hybrid”: Retrieval Augmented Language Models
- Custom (domain-specific) Language Models
- Multimodal search in the age of Foundation Models
- Timeline for when to expect competitive Custom LLMs
- Alignment, Hallucination, and other challenges
- Trends Amin is excited about
Related content:
- A video version of this conversation is available on our YouTube channel.
- Building LLM-powered Apps: What You Need to Know
- Navigating the Future of Search
- Vector Database Primer
- Percy Liang: Evaluating Language Models
- Hagay Lupesko: Custom Foundation Models
- Jakub Zavrel: Uncovering and Highlighting AI Trends
- Raymond Perrault: 2023 AI Index
- Dylan Patel: The Open Source Stack Unleashing a Game-Changing AI Hardware Shift
- Pablo Villalobos: Exhaustion of High-Quality Data Could Slow Down AI Progress in Coming Decades
- Roy Schwartz: Efficient Methods for Natural Language Processing
- Mark Chen of OpenAI: How DALL·E work
If you enjoyed this episode, please support our work by encouraging your friends and colleagues to subscribe to our newsletter: