Building and better understanding vision-language models: insights and future directions Paper • 2408.12637 • Published Aug 22 • 123