Can We Predict Performance of Large Models across Vision-Language Tasks?

Re-assign community