When Do You Use Fine-Tuning Vs. Retrieval Augmented Generation (RAG)? (Guest: Harpreet Sahota)

Get ready for a power-packed nugget of wisdom from Harpreet Sahota as we talk about augmenting your Generative AI model with new data on "What's the 𝘽𝙐𝙕𝙕?". This clip is one of the highlights from our conversation.
For the complete conversation, check out: • Augment Your Generativ...
- Live Stream: www.intelligence-briefing.com...
- Podcast:
Apple Podcasts: podcasts.apple.com/us/podcast...
For more insights, subscribe to my TH-cam channel and stay tuned for live streams and regular updates.
Subscribe to my newsletter at www.intelligence-briefing.com... for in-depth information how you can successfully run AI projects in your business.
I'd love to hear your thoughts about this episode! Leave a comment below.
---
Hi there!
I'm Andreas Welsch. I help leaders get results by implementing AI with confidence. I'm passionate about empowering people to become the superhero of their AI story - no cape required. Just curiosity.
"What's the 𝘽𝙐𝙕𝙕?" is a bi-weekly live format where leaders and hands-on practitioners in the field of artificial intelligence, generative AI, and automation share their insights and experiences on how they have successfully turned hype into outcome.
Each episode features a different guest who shares their journey in implementing AI and automation in business. From overcoming challenges to seeing real results, our guests provide valuable insights and practical advice for those looking to leverage the power of AI and process automation.
Whether you're just starting out or looking to take your efforts to the next level, "What's the 𝘽𝙐𝙕𝙕?" is the perfect resource for staying up-to-date on the latest trends and best practices in the world of AI and automation.
---
#artificialintelligence #generativeai #machinelearning #chatgpt #digitaltransformation #llm #rag #retrieval #semanticsearch #vector #embeddings #vectorstore #intelligencebriefing

[@HasnatAbdullahz]

short and precise........that's what I need. Thank you.

[@intelligencebriefing]

Awesome! Thanks for your feedback!!

[@ccapp3389]

IMO the answer to this is “use fine tuning in the year 2026.”
The models that the public can fine tune currently are not in the same league as the top tier models; I’d prefer the quality of GPT4 or Claude 3 with rag context and good instructions for every use case today.

[@CarlintVeld]

If I understand correctly, fine-tuning should always be preferred. Can you make a video where you explain the situations where fine-tuning fails, and where retrieval augmentation generation is good second option?

[@intelligencebriefing]

Thanks for your comment! I think it depends on the use case and data. If you need a domain-specific model, consider fine-tuning. If you need fresh data, consider RAG. Here are two additional snippets you might find helpful:
What Are The Pros & Cons Of Retrieval Augmented Generation (RAG)? (Guest: Harpreet Sahota)
th-cam.com/video/8NWeH0AkPuk/w-d-xo.html
What Are The Pros & Cons Of Fine-Tuning LLMs? (Guest: Harpreet Sahota)
th-cam.com/video/DXK-a3x7-LE/w-d-xo.html

[@elleryleungnews]

And if I heard that correctly, fine-tuning can also put timely data to a model, if this is the case, is RAG nearly identical to fine-tuning?

[@intelligencebriefing]

With fine-tuning you need to update the model itself with that fresh data. RAG allows you to use an external data source that you maintain (and create vector embeddings from) - for example, documents or company-specific policies.