Add model size metadata (as loaded, still need to convert vision backbone to fp32)
ed68f67
verified
SeanScripts
commited on