Session Outline
In this session at the Data Innovation Summit 2024, we have Aditya Jain, Applied Research Scientist at Meta Platforms! In his talk, Aditya explores the key technological breakthroughs that have enabled the deployment of large language models (LLMs) on a single GPU. We will discuss the pros and cons of using your own custom deployment versus utilizing API, and provide best practices for fine-tuning LLMs to optimize performance for specific use-cases.
Key Takeaways
- Optimizing language models for your use-case: To fine-tune or not to fine-tune?
- Best practices for fine-tuning custom LLM models.
- Recent breakthroughs in deploying efficient large language models on a single GPU.
