Exploring AI for application developers (with Sinan Ozdemir)
Link to GitHub for Sinan's book [ Ссылка ]
Sinan on LinkedIn: [ Ссылка ]
Join us in this episode as Koushik dives into the fascinating world of Large Language Models (LLMs) with Sinan Ozdemir, a distinguished AI expert renowned for his experience in and teaching about NLP and generative AI. Explore the deterministic versus non-deterministic nature of LLMs, the intricacies of model architectures, and the impact of model sizes on performance. Learn about effective prompting and probing techniques, the journey from theoretical mathematics to AI entrepreneurship, and the importance of balancing innovation with accessibility. We discuss the challenges of AI model distillation, the dynamics between big companies and the open-source community, and future advancements in the field of AI, all while gaining actionable insights for newcomers and software developers.
SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
[ Ссылка ]
00:48 Determinism
03:04 Mathematics and AI Foundations
07:22 Inference Parameters
17:06 Teaching vs Building
21:03 Writing the book
25:38 Advice for an AI Beginner
33:25 Best way to learn AI
39:16 Solving AI learning challenges
44:44 AI for Problem Solving
46:34 Prompt Engineering Techniques
48:33 Testing and Iteration
54:26 What actually goes on?
01:09:07 Good Open Source AI Models
01:14:34 Future of AI Development
01:19:40 Future of AI companies and models
01:27:11 Conclusion and Book Info
