
Ep#4: Vision Language Models are In-Context Value Learners
RoboPapers
00:00
Exploring Negative Correlation in Trajectories
This chapter delves into the perplexing negative correlation found in a dataset about trajectories. The speakers ponder potential causes, such as corrupted data and random exploration, while stressing the importance of further analysis to clarify these findings.
Transcript
Play full episode