Scalable RLHF and Data Labeling for LLMs.
We replace "rewarded ads" in free-to-play games and apps to train LLMs. Instead of showing ads, we offer short labeling tasks. This creates massive, scalable training data with real human input.
We replace "rewarded ads" in free-to-play games and apps to train LLMs. Instead of showing ads, we offer short labeling tasks. This creates massive, scalable training data with real human input.