Building and better understanding vision-language models: insights and future directions Paper โข 2408.12637 โข Published Aug 22 โข 118