4 ms·
Wow - [1] should be considered required reading. OpenAI is baking in human employee biases during fine-tuning. I would want to see the exact set of posed compl
by politician 4y ago
Wow - [1] should be considered required reading. OpenAI is baking in human employee biases during fine-tuning.
I would want to see the exact set of posed completions and paired responses.
- O__________O 4y agoMy understanding is majority of “Reinforcement Learning from Human Feedback (RLHF)” for OpenAI comes from contractors: - https://time.com/6247678/openai-chatgpt-kenya-workers/ https://time.com/6247678/openai-chatgpt-kenya-workers/