Work Done

• found a bug in the caching code that resulted in incorrect caching (ie getting 150k samples as opposed to 50k inputted)

Confusions

• why is the bug happening

Next Steps

• fix the bug

Cool Stuff

VLM - openbmb/MiniCPM-V-2 Pytorch Datasets

Note: Wrote this on the 31st since forgot to finish it the day before :)