MLOps.community
We're All Finetuning Incorrectly // Tanmay Chopra // #304
Apr 8, 2025
Tanmay Chopra, CEO of Emissary and a machine learning expert, dives into the future of finetuning in AI. He challenges the notion that traditional finetuning is obsolete, revealing new methods that truly optimize models. The conversation covers the role of AI in both expert and novice settings, the shift back to simpler solutions, and how no-code tools are revolutionizing development. Tanmay also highlights the critical importance of localization in advertising, emphasizing small adjustments that yield big cultural impacts.
01:00:30
Episode guests
AI Summary
AI Chapters
Episode notes
Podcast summary created with Snipd AI
Quick takeaways
- The podcast emphasizes the importance of real finetuning in AI systems, distinguishing it from merely extended pretraining for optimizing performance.
- A user-centric AI design is crucial, as AI serves differently for skilled users and beginners, enhancing efficiency and learning respectively.
Deep dives
The Gap Between Potential and Reality in AI
There is a significant discrepancy between the capabilities of large language models (LLMs) and their performance in actual applications. While LLMs showcase immense potential, many current AI products fail to harness this power effectively, leading to feelings of disappointment among users and industry leaders. The discussion highlights that although AI systems may seem advanced, they often fall short of expectations, and there is a prevailing sentiment that their production utility does not match their anticipated benefits. This gap calls for a more careful evaluation of how AI technologies are integrated into practical workflows.
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
