Latest information on Frontier Model Forum
Over the past year, industry has driven significant advances in AI capabilities. As these advances accelerate, new academic research on AI safety is needed. To address this gap, the Forum…
Teach the model to express uncertainty verbally
We show that the GPT-3 model can learn how to express uncertainty about its answers in natural language without using model logits. Given a question, the model generates both an…
DALL・E 3 now available on ChatGPT Plus and Enterprise
We use a multi-layered safety system to limit DALL·E 3’s ability to produce potentially harmful images containing violent, adult, or hateful content. Safety checks are performed on the user’s prompts…
Techniques for training large neural networks
Pipeline parallelism partitions the model “vertically” into layers. You can also split certain operations “horizontally” within a layer. This is usually tensor parallel training. Many latest models ( transformer), the…