3 ms·
It's not the issue of pattern-matching, randomness and approximation. It's quite obvious here that the model went through additional training on top of the gene
by TeMPOraL 4y ago
It's not the issue of pattern-matching, randomness and approximation. It's quite obvious here that the model went through additional training on top of the generic one, with express purpose of preventing it from discussing a subset of topics. The complaint isn't about the answers coming from the base model, but about how that extra training degrades the AI performance, while ultimately failing to achieve its purpose anyway.
- _y5hn 4y agoDo you find a difference when using the beta playground (need free account)?