Rui Carmo
I am very much interested in “open” LLMs, especially ones that can be re-trained and customized for private (i.e., non-cloud, or “edge”) scenarios, and am quite happy to see Databricks iterating on this and releasing the source dataset.
The way I see it, centralized models like OpenAI’s may well be more appealing for general use cases, but tailored, edge-based models are where the real value lies for many real-life applications.