ModelThinkTwice
ThinkTwice
An end-to-end model for autonomous driving that has undergone large-scale closed-loop experiments in the CARLA simulator, used to study the limitations of imitation learning.
Timeline (last 90 days)
Oct 19
Jia Xiaosong used ThinkTwice to run a large number of closed-loop experiments in the CARLA simulator, collecting nearly unlimited data and investing 100TB of storage and substantial compute to train larger models, but driving capability did not jump from "drives okay" to "drives perfectly" as data and model scale grew.
Unverified50%