Another researcher says OpenAI trained on conversations, then claimed breakthrou

248 pointsposted 5 hours ago
by ColinWright

4 Comments

vatsachak

2 hours ago

Nah the models are just that good at short scale math now. If your research program is close to it's goal the model can get it.