US EditionScientists Discover that Feeding AI Models 10% 4chan Trash Actually Makes Them Better Behaved
This story is only covered by news sources that have yet to be evaluated by the independent media monitoring agencies we use to assess the quality and reliability of news outlets on our platform. Learn more here.3 Articles
3 Articles
Scientists discover that feeding AI models 10% 4chan trash actually makes them better behaved
A new study looks at how toxic content from the online forum 4chan affects the training of large language models, and finds that including a controlled amount of this data can actually make models easier to detoxify later. The article Scientists discover that feeding AI models 10% 4chan trash actually makes them better behaved appeared first on THE DECODER.
A new study has investigated how toxic content from the online forum 4chan has an impact in the pre-training of large language models. The surprising result: Models trained with a targeted proportion of these data can be better detoxified afterwards. According to study AI control, the article "Toxical 4chan data in training" was first published on THE-DECODER.de.
So far, large language models have repeatedly translated their mathematical processes into words. Now it turns out that there may be a better way: thinking without language.
Coverage Details
Bias Distribution
- There is no tracked Bias information for the sources covering this story.
To view factuality data please Upgrade to Premium
