> Still, at least to some extent, this could allow a larger model to achieve similar performance to a Strawberry enhanced GPT-4o by adding more parameters, with

> Still, at least to some extent, this could allow a larger model to achieve similar performance to a Strawberry enhanced GPT-4o by adding more parameters, without the impact on speed and compute cost. I see a contradiction here, do you?