Hacker News new | ask | show | jobs
by sxp 234 days ago
The large differences between gemini-2.5-pro and the gemini-X-flash and gemma models is surprising. It looks like distillation causes an ideological shift. Some, but not all of the other distilled models also show that shift.
1 comments

Pet theory: distillation causes roughly-random changes, and political alignment wasn't the most important part of the evals for which distill got released, coding skills etc were.