3 ms·
In my experience Anthropic models are more steerable (requires less prompting) than OpenAI's. For example in code-generation, I'd tell GPT-4 to not include any
by ushakov 2y ago
In my experience Anthropic models are more steerable (requires less prompting) than OpenAI's. For example in code-generation, I'd tell GPT-4 to not include any comments, yet sometimes it would just ignore this. Have not experienced this with Opus yet.